Uncategorized

Summary

This panel discussion focused on the intersection of artificial intelligence and environmental sustainability, exploring how AI can both contribute to and help solve climate challenges. The moderator introduced the central paradox that while AI offers opportunities to address environmental risks, it simultaneously contributes to the problem through high energy and water consumption.

Mario Nobile from Italy’s digital agency outlined his country’s four-pillar AI strategy emphasizing education, research, public administration, and enterprise applications. He highlighted efforts to transition from energy-intensive large models to smaller, vertical domain-specific models for sectors like manufacturing, health, and tourism. Marco Zennaro presented TinyML technology, which runs machine learning on extremely small, low-power devices costing under $10, enabling AI applications in remote areas without internet connectivity. He shared examples from a global network of 60 universities developing applications like disease detection in livestock and anemia screening in remote villages.

Adham Abouzied from Boston Consulting Group emphasized the value of open-source solutions, citing a Harvard study showing that recreating existing open-source intellectual property would cost $4 billion, but licensing it separately would cost $8 trillion. He advocated for system-level cooperation and data sharing across value chains to optimize AI’s impact on sectors like energy. Ioanna Ntinou discussed her work on the Raido project, demonstrating how knowledge distillation techniques reduced model parameters by 60% while maintaining accuracy in energy forecasting applications.

Mark Gachara from Mozilla Foundation highlighted the importance of measuring environmental impact, referencing projects like Code Carbon that help developers optimize their code’s energy consumption. The panelists agreed that sustainable AI requires active incentivization rather than emerging by default, emphasizing the need for transparency in energy reporting, open-source collaboration, and policies that promote smaller, task-specific models over large general-purpose ones. The discussion concluded that achieving sustainable AI requires coordinated efforts across education, policy, procurement, and international cooperation to balance AI’s benefits with its environmental costs.

Keypoints

## Major Discussion Points:

– **AI’s dual role in climate challenges**: The discussion explored how AI presents both opportunities to address environmental risks (through optimization and efficiency) and contributes to the problem through high energy and water consumption, creating a complex “foster opportunity while mitigating risk” scenario.

– **Small-scale and efficient AI solutions**: Multiple panelists emphasized moving away from large, energy-intensive models toward smaller, task-specific AI applications, including TinyML devices that cost under $10, consume minimal power, and can operate without internet connectivity in remote areas.

– **Open source collaboration and data sharing**: The conversation highlighted how open source approaches can significantly reduce costs and energy consumption by avoiding duplication of effort, with examples showing potential savings from $4 billion to $8 trillion through shared intellectual property.

– **Policy frameworks and governance models**: Panelists discussed the need for comprehensive governance including procurement policies, regulatory frameworks, transparency requirements, and incentive structures (both “carrots and sticks”) to promote sustainable AI development and deployment.

– **Transparency and measurement imperatives**: A key theme was the critical need for transparency in AI energy consumption, with calls for standardized reporting of energy use at the prompt level and better assessment tools to enable informed decision-making by developers, policymakers, and users.

## Overall Purpose:

The discussion aimed to explore practical solutions for balancing AI’s potential benefits in addressing climate change with its environmental costs, focusing on how different stakeholders (governments, researchers, civil society, and industry) can collaborate to develop and implement more sustainable AI technologies and governance frameworks.

## Overall Tone:

The discussion maintained a consistently optimistic and solution-oriented tone throughout. While acknowledging the serious challenges posed by AI’s environmental impact, panelists focused on concrete examples of successful implementations, practical policy recommendations, and collaborative approaches. The moderator set an upbeat tone from the beginning by emphasizing hope and “walking the talk” rather than dwelling on problems, and this constructive atmosphere persisted as panelists shared specific use cases, technical solutions, and policy frameworks that are already showing positive results.

Speakers

– **Mario Nobile** – Director General of the Agents for Digital Italy (AGIDS)

– **Leona Verdadero** – UNESCO colleague, in charge of the report “Smarter, Smaller, Stronger, Resource Efficient Generative AI in the Future of Digital Transformation”

– **Marco Zennaro** – Next Edge AI Expert at the Abdus Salam International Center of Theoretical Physics, also works for UNESCO, works extensively on TinyML and energy efficient AI applications

– **Audience** – Jan Lublinski from DW Academy in Media Development under Deutsche Welle, former science journalist

– **Adham Abouzied** – Managing Director and Partner at Boston Consulting Group, works at the intersection of AI, climate resilience and digital innovation with focus on open source AI solutions

– **Moderator** – Panel moderator (role/title not specified)

– **Ioanna Ntinou** – Postdoctoral researcher in computer vision and machine learning at Queen Mary University of London, works with Raido project focusing on reliable and optimized AI

– **Mark Gachara** – Senior advisor, Global Public Policy Engagements at the Mozilla Foundation

Additional speakers:

None identified beyond the provided speakers names list.

# AI and Environmental Sustainability Panel Discussion Report

## Introduction

This panel discussion examined the intersection between artificial intelligence and environmental sustainability, addressing what the moderator described as a central paradox: while AI offers opportunities to tackle climate challenges, it simultaneously contributes to environmental problems through substantial energy and water consumption. The moderator emphasized the need to move beyond theoretical discussions toward practical solutions, setting a solution-oriented tone for the conversation.

The panel brought together perspectives from government policy, academia, civil society, and industry to explore how different stakeholders can collaborate on more sustainable AI technologies and governance frameworks.

## National Strategy and Policy Frameworks

### Italy’s Four-Pillar Approach

Mario Nobile, Director General of Italy’s Agency for Digital Italy (AGID), outlined Italy’s comprehensive AI governance strategy built on four pillars: education, scientific research, public administration, and enterprise applications. He emphasized that the contemporary debate has evolved, stating: “Now the debate is not humans versus machines. Now the debate is about who understands and uses managed AI versus who don’t.”

Nobile highlighted Italy’s substantial financial commitment, with 69 billion euros allocated for ecological transition and 13 billion euros for business digitalization from the National Recovery Plan. He noted the challenge of implementing guidelines across Italy’s 23,000 public administrations and described plans for tax credit frameworks to incentivize small and medium enterprises to adopt AI technologies.

A key element of Italy’s strategy involves transitioning from large, energy-intensive models to what Nobile called “vertical and agile foundation models” designed for specific sectors including manufacturing, health, transportation, and tourism. Italy’s Open Innovation Framework enables new procurement methods beyond conventional tenders for public administration, addressing regulatory challenges around data governance and cloud usage.

## Technical Solutions and Innovations

### TinyML: Small-Scale AI Applications

Marco Zennaro from the Abdus Salam International Centre of Theoretical Physics presented TinyML (Tiny Machine Learning) as an approach to sustainable AI. This technology enables machine learning on small, low-power devices costing under $10, consuming minimal energy, and operating without internet connectivity.

Zennaro described a global network of over 60 universities across 32 countries developing practical TinyML applications, including:

– Disease detection in cows in Zimbabwe

– Bee population monitoring in Kenya

– Anemia screening in remote villages in Peru

– Turtle behavior analysis for conservation in Argentina

He posed a fundamental question: “Do we always need these super wide models that can answer every question we have? Or is it better to focus on models that solve specific issues which are useful for SDGs or for humanity in general?”

The TinyML approach emphasizes regional collaboration and south-to-south knowledge sharing, with investment in local capacity building using open curricula developed with global partners.

### Model Optimization Research

Ioanna Ntinou from Queen Mary University of London presented research from the Radio project demonstrating how knowledge distillation techniques can achieve significant efficiency gains. Her work showed that model parameters could be reduced by 60% while maintaining accuracy in energy forecasting applications.

Ntinou raised concerns about current evaluation frameworks, noting: “If we are measuring everything by accuracy, and sometimes we neglect the cost that comes with accuracy, we might consume way more energy than what is actually needed.” She emphasized that sustainable AI development requires active intervention and new success metrics that balance performance with environmental impact.

## Open Source Collaboration and Economic Perspectives

### The Economics of Shared Development

Adham Abouzied from Boston Consulting Group presented economic evidence for open source collaboration, citing research showing that recreating existing open source intellectual property would cost $4 billion, while licensing equivalent proprietary solutions separately would cost $8 trillion. He argued that “open source solutions optimize energy costs by avoiding repetitive development.”

Abouzied emphasized that meaningful AI applications in sectors like energy require system-level cooperation and data sharing across value chains, noting that “AI breakthroughs came from training on wealth of internet data; for vertical impact, models need to see and train on data across value chains.”

## Civil Society and Environmental Justice

### Community-Centered Approaches

Mark Gachara from Mozilla Foundation brought environmental justice perspectives to the discussion, emphasizing that “the theater of where the most impact of climate is in the global south and it would be a farmer, it would be indigenous and local communities.”

Gachara shared a specific example from Kilifi, Kenya, where communities used AI tools for ecological mapping to advocate against a proposed nuclear reactor project. He highlighted how civil society can leverage AI for evidence generation while ensuring that communities most affected by climate change have agency in developing responses.

He referenced tools like Code Carbon that help developers measure their code’s energy consumption, demonstrating how transparency can drive behavioral change at the individual developer level.

## Transparency and Measurement Challenges

### The Need for Energy Reporting

A recurring theme was the urgent need for transparency in AI energy consumption reporting. Ntinou highlighted that the current lack of visibility into energy usage per prompt or model interaction makes it impossible to develop effective legislation or make informed decisions about AI deployment.

The panelists agreed that assessment of energy usage in widely used public models is essential before focusing on regulatory frameworks. Without standardized evaluation methods and transparent reporting, stakeholders cannot compare the environmental impact of different AI systems.

An audience question from Jan Lublinski raised the possibility of carbon trading mechanisms for AI, asking whether transparency in energy consumption could enable market-based solutions for reducing AI’s environmental impact.

## UNESCO Report Preview

Leona Verdadero participated online to provide a preview of UNESCO’s forthcoming report on resource-efficient generative AI. While details were limited in this discussion, she indicated the report would provide evidence about optimized AI models and their potential for improved efficiency.

## Key Themes and Future Directions

### Emerging Consensus

The discussion revealed broad agreement across speakers on several key principles:

– The value of smaller, task-specific AI models over large general-purpose systems

– The critical importance of transparency in energy consumption measurement

– The benefits of open source and collaborative approaches

– The need for comprehensive policy frameworks that balance innovation with sustainability

### Implementation Challenges

Several challenges remain unresolved, including the fundamental tension between AI model accuracy and energy consumption when accuracy remains the primary success metric. Regulatory challenges around data governance and cross-border data transfer continue to limit AI implementation, particularly in emerging economies.

The lack of transparency in energy consumption reporting for widely used AI models represents a critical gap that must be addressed before effective policy interventions can be developed.

## Conclusion

The discussion demonstrated that achieving sustainable AI requires coordinated efforts across education, policy, procurement, and international cooperation. The panelists agreed that sustainable AI will not emerge by default and requires active incentivization through both regulatory frameworks and market mechanisms.

The conversation revealed convergence toward practical, implementable solutions emphasizing efficiency over scale, transparency in measurement, open collaboration, and equitable access. The path forward involves not just technical innovation but fundamental changes in how success is measured and resources are allocated in AI development.

As the discussion highlighted, the solutions exist across multiple scales—from tiny, efficient models addressing specific local challenges to policy frameworks enabling sustainable AI adoption at national levels. What remains is coordinated implementation at the scale and speed required to address both AI’s environmental impact and its potential to help solve climate challenges.

Moderator: Good morning, everyone. A pleasure to be here with you all, in the audience, online, but also with this fantastic group of panelists. I’m sure if you read any of these reports about the most important issues of our times, you will certainly find the issues that we are discussing here today. Some reports will say governance of artificial intelligence is one of the most important issues we need to deal with. Certainly, climate change, many of these reports will say it’s one of the most important policy issues we have in front of us, as well as it’s one of the most threatening risks for humanity. Energy resources and scarce resources, and water is probably another of those issues that you’ll see in different reports here and there. The challenge for this gentleman and lady here is how to combine all those things together. And of course, with you, I hope we are going to have a very interesting conversation about that. So, I don’t want to use a cliché here, but voilà, artificial intelligence offers, maybe needless to say, interesting opportunities for addressing some of these environmental risks, and we can speak about that, how the models can help us, and us as the climate scientists, as the policy makers, as the civil society, to solve this very complex equation on how we solve the planetary crisis that we are in the middle of, but tragically, or ironically, the same technology that can help us to address the issue is contributing to the problem, because it consumes a lot of energy, and water, and so on. So, again, as I said, it’s a cliché, but our job is how we can foster the opportunity and mitigate the risk. And as usual, in these very complex lives we have right now, it’s easy to say, but not necessarily easy to do. The good news is that these people in this panel and online, they do have some interesting solutions to propose to you, and some of them are already implementing it. So, since I’m optimistic by nature, when my team asked me to moderate and I saw what they prepared, I was very happy that it was not only dark and terrible, but it was also about how we can actually walk the talk, right? So, this session is a lot about this, to have a dialogue with this group of different actors here that are dedicated to think about this problem, and how we can… and I think we can suggest and underline in the one hour we have some of these issues. Of course, let’s look into the implications of AI technologies for these problems. We will try to showcase some tools and frameworks. UNESCO with the University College of London, we are going to launch very soon a very exciting issue brief called Smarter, Smaller, Stronger, Resource Efficient Generative AI in the Future of Digital Transformation that my dear colleague Leona, who is online, unfortunately she couldn’t be here today also to reduce UNESCO’s carbon footprint of people travelling all over the world, but she is the person in charge of this report and you can ask further questions and interact with her on that. And also again, as I said, try to highlight the different approaches that we can take to address these issues. So I’m sure we are going to have an exciting panel and since we need to be expeditive, I will stop here and go straight to my panellists and let me start with you, Mario. So Mario Nobile is the Director General of the Agents for Digital Italy, AGIDS, I guess I say like that. Correct. And Mario, let’s first start on what you are already doing at the strategic level, right? So how your agency is trying to cope with these not easy challenges. Buongiorno, over to you.

Mario Nobile: Buongiorno, thank you and good morning to all. Our Italian strategy rests on four pillars. Education, first of all. Scientific Research, Public Administration and Enterprises. And these efforts aim to bridge the divide, ensuring inclusive growth and empowering individuals to thrive in an AI-driven economy. For the first, scientific research, our goal is promoting research and development of advanced AI technologies. You before mentioned technology evolves really fast. So now we are dealing with agentic AI. We started with large language models, then large multimodal models. Now there is a new frontier about small models for vertical domains. So we met also with Confindustria, which is the enterprises organization in Italy. And we are trying applications about manufacturing, health, tourism, transportation. And this is important for energy-consuming models. For public administration, we are trying to improve the efficiency and effectiveness of public services. For companies, Italy is the seventh country in the world for exports. So we must find a way to get to the application layer and to find concrete solutions for our enterprises. And education, first of all. I always say that now the debate is not humans versus machines. Now the debate is about who understands and uses managed AI versus machines. who don’t. Okay. And in Italy, the AI strategy is, we wrote it in 2024, we have also a strategic planning, the three year plan for IT in public administration. And we emphasize the importance of boosting digital transition using AI in an ethical and inclusive way. This is important for us. We have three minutes, so I go to the conclusion. And we are stressing our universities about techniques like incremental and federated learning to reduce model size and computational resource demands. This is the first goal. And so this approach minimizes energy consumption and we are creating the conditions to transition from brute force models, large and energy consuming, to vertical and agile foundation models with specific purposes, health, transportation, tourism, manufacturing. This is the point. Now, I was saying technology evolves faster than a strategy. So we have a strategy, but we are dealing with agentic AI, which is another frontier.

Moderator: Thank you very much, Mario. And I’m glad to hear the conclusion in terms of what you are working with your universities. Because I mean, I’m just the international bureaucrat here, right? I don’t understand anything about these things. But I did read the report that my team prepared and they were making recommendations towards what you were saying. And one of my questions to them is, is this Marco Zennaro, Next Edge AI Expert at the Abdus Salam International Center of Theoretical Physics. He also works for UNESCO. My children, they always tell me, why don’t you do these kind of interesting things? Because they don’t understand what I do, right? Because maybe not even I, but people like Marco are the ones walking the talk in UNESCO. So Marco, you have worked extensively on 10 ml and energy efficient AI applications, also in African contexts. So can you tell us, and especially for a person like me that is no political scientist, we don’t understand anything. So how you can understand the benefits of what you are doing?

Marco Zennaro: Sure, sure. Definitely. Thank you very much. So let me introduce TinyML first. So TinyML is about running machine learning models on really tiny devices, on really small devices. And when I say small, I say really small. So devices that have, you know, few kilobytes of memory, that have really slow processors, but they have two main advantages. The first one is that they’re extremely low power. So we’re talking about, you know, green AI, these devices consume very little power. And second advantage is that they’re extremely low cost. So one of these chips is less than a dollar, and one full device is about $10. So, you know, that’s, of course, very positive. And they allow AI or machine learning to run on the devices without the need of internet connection. So we heard, you know, during IGF that a third of the world is not connected. And if we want to have, you know, data from places that are remote, where there’s no internet connection, and we want to use AI or machine learning, that is a really good solution. So you ask about application. And so in 2020, together with colleagues from Harvard University and Columbia University, we created a network of people working on TinyML with a special focus on the Global South. Now we have more than 60 universities in 32 different countries. So we have many researchers working on TinyML in different environments. And they worked on very diverse applications. So just to cite a few, there’s colleagues in Zimbabwe that worked on TinyML for food and mouth disease in cows. So, you know, sticking this device in the mouth of the cow and detecting the disease. There’s colleagues in Kenya working on counting the number of bees that you have in a beehive. There’s colleagues in Peru that use TinyML to detect anemia through the eyes in remote villages. We have colleagues in Argentina that use TinyML on turtles to understand how they behave. So very diverse applications. Many of them have impact on SDGs. And again, using low cost and extremely low power devices.

Moderator: Thank you. This is fascinating. As you said, we are having in this IGF, which I think is a good thing, lots of discussions regarding, I mean, in this UN language, how we do these things, leaving no one behind. And these are very concrete examples, right? Because it’s about the cost, it’s about being low intensive on energy. So very glad to hear that. And again, to see that those things are possible. And this offers us a bit of hope, because as I said in the beginning, I’m always concerned that we are only looking to the side of the problems and the terrible risks, but without looking into what is already happening to address the issues, right? So let me move now to have a different setting to our online guests. We have One speaker that is speaking from the internet space, from the digital world. And he is Adham Abduzadeh, and he is Managing Director and Partner at the Boston Consulting Group. Welcome, Adham, to this conversation. And I know that the BCG, the Boston Consulting Group, is very much concerned about these issues as well. You are putting a lot of effort on that. And you yourself have worked at the intersection of AI, climate resilience and digital innovation. And with a specific focus on the open source AI solutions. So how you can tell us in your three minutes the connection of these issues with the main topic of this panel, that of course is the relationship with sustainability, environmental sustainability. Welcome and over to you.

Adham Abouzied: Thank you very much. Honored and very happy to join this very interesting panel. And I must say I enjoyed so much the interventions of the panelists before me. I think you asked a very, very interesting question. I will start my answer by basically the results of a study that has been recently made by Harvard University, which is around the value of the open source wealth that exists on the ground today. The study basically estimates that if we would recreate only one time the wealth of open source intellectual property that is available today on the Internet, it would cost us $4 billion. But it does not stop there, because if you assume or if you imagine that this material. was not open source, and that every player who would want to use it would either recreate it or pay a license, then basically you would increase this $4 billion of cost to $8 trillion just because of the repetition. So in a sense, having something that is open source optimizes significantly the amount of work that is required to get the value out of these AI algorithms or digital technologies in a larger sense. So instead of doing the thing one time, you will have communities that are actually contributing and building on top of the, I would say, intellectual creation of each other. First, I mean, to be able to get to broader impact at a much lower cost and a much lower energy cost as well. Now, from experience also, and building specifically on what Mario is saying, for you to be able to implement vertically focused AI models that generate value across a certain system or a certain value chain, you need to have some kind of system level cooperation, which means that today all of the, I would say, AI breakthroughs that we have been seeing basically through very large, generic foundational models, it’s there because they were trained on the data that is available out there on the internet, the wealth of text, the wealth of images, the wealth of videos, and basically these models are good at this generic, generative, tasks because of what these models have been able to see and train on. But now for you to be able to have a meaningful impact with models that have a vertical focus and can create value across a certain system, they have to basically see and train on data and build on top of the decisions that are made across a value chain. Let me give a concrete example that relates to this basically your starting speech. Let’s take the energy sector today. If you want to have or if you’re seeking to have basically an AI development that rather than creating a burden on the energy system, actually it’s becoming more optimised, then you also need to allow AI to create value within the energy sector, basically help optimise the decisions from the generation to the transmission to the actual usage. And if I take this energy system as an example, basically currently in very few countries in the world, you can see basically cooperation, data exchange, IP exchange across the different steps of the value chain. This would be essential and if it is allowed through policies, through protocols, like when we first started the internet with TCP IP, then the vertical models can create much more significant value that would allow, as a big example, the energy sector to optimise its income and will allow the AI algorithms not only to consume less energy by themselves, but also to help the energy sector itself. I would say, optimise its output and become greener.

Moderator: Thank you. Very interesting. So at the end of the day, we need to move from the lose-lose game to the win-win game, right? And you mentioned some keywords that I’m sure Jona will talk about, optimisation, cooperation, so that’s a very interesting segue. Let me just make one remark on the open source and the openness, on the open solutions, that is very interesting. We witnessed the concrete global example during the pandemic, right? When the scientists decided to open, the vaccines were produced in a record time in human history. So here, mutatis, mutandis, we need to use the same logic to fix this huge problem. So, Jona, in TINU, it’s correct? It’s TINU. OK, sorry about that. So Jona is a postdoctoral researcher in computer vision and machine learning at the Queen Mary University in London. And you work with Raido project, which focuses on reliable and optimised AI, again, the word. So, again, can you walk us through the specific use case from the project where you managed to show this connection with the efficiency and so on? Over to you.

Ioanna Ntinou: Yes, thank you. I’m happy to be here. So, as you said, Raido stands for reliable AI in data optimisation. And what we’re trying to do is to be a bit more conscious when we develop AI models on the energy consumption that the models are going to use. And by the end of the project, we will have developed a platform that we are going to test it against for real-life use cases. We call it pilots. And that they are spun in different domains. This can be… Institute of Technology, Robotics, Healthcare, Smart Farming, and Critical Infrastructure. Today, I will focus in one specific domain, which is the energy grid, and we’re working with two companies from Greece. One is an energy company, the National Energy Company, which is called PPC, and the other is like a research center called CERF. So, what we have is an optimized model for day ahead forecasting of energy demand and supply, particularly in smart homes that they have a microgrid. And what we were given was a big time series model that would predict the energy demand of small electronic devices that we have in our house, including, let’s say, a bulb. The problem with this model was that it was a bit big. It was very good in accuracy, but it was quite big, and it would need quite often retraining because, as you know, when we calculate the energy consumption of a device, this is dependent on the seasonality, this is dependent on the different habits that someone can have in a house. And what we did was simply a knowledge distillation. We took this model and we used it as a teacher to train a smaller model to produce more or less the same results. It was quite close in accuracy, but we reduced the number of parameters to 60%. So, we got a much smaller model, we call it the student now, that has more or less the same performance, and we got several benefits from this process. First, we have a smaller model that consumes much smaller energy. Secondly, this model is easier to deploy in small devices, as we discussed before in the panel, right, which is very critical because then you can democratize the… access to AI to a lot of houses, right? Because if you have a small model, you can deploy it much easier. And also, by maintaining the accuracy, we somehow help these companies have a more accurate forecast of the energy demand that they will be around. And in that way, they can consume or they produce energy having this into their mind. So we are happy with this and we hope that we will help also the rest of the pilots get good models and have good results.

Moderator: Thank you. Fascinating. And so it’s also a bit about evidence-based next steps, either for the companies or for policymaking. I remember a few years ago, before the AI times, but already in the open data environment, a Ministry of Energy of a particular country launched a hackathon with open data so that people could help to improve the efficiency of energy consumption and so on in the public sector. So transparency always sheds light, sometimes in very funny ways. And after they did the hackathon and the people did the run the models and so on, they found out that the ministry that consumed more energy in that country was the Ministry of Energy and Environment, which was a big shame for them. But then they actually implemented some very concrete policies to resolve these issues. So that’s also interesting. We need to connect these conversations with the conversation of transparency and accountability and so on. So, Mark, last but not least, Mark Gachara is a senior advisor, Global Public Policy Engagements at the Mozilla Foundation. And we are going to get back to the conversation, I guess, of open source because Mozilla has a long experience on that. And this mission of Mozilla

Mark Gachara: We had a number of grantees think about this particular area. We had, for example, an organization from France called Code Carbon. And within this theme of environmental justice, they were thinking about, when I write code, how much energy am I using? Because at the end of the day, fossil fuels are used to generate the energy. So a developer can actually optimize their code. And it’s an open source project that’s available that people can. before they harm the environment. These are just examples of practical use cases that are measuring because in management they have told us if you can’t measure it, you can’t manage it. Probably you also can’t manage the risk. So I think being able to run this kind of action research that can talk to policy eventually then sheds light to make like environmental justice part of a core definition of how we roll out AI solutions.

Moderator: Thank you. Thank you. Super interesting and if I hear you correctly there are two important issues. There may be a energy efficiency AI by design right from the moment you are writing the code which is an interesting story also from the human rights perspective to be by design but also the capacity of having decent risk assessments for these things. So let’s now what’s going to happen we are going to have a second round here very quickly ping pong and then we are going to open to you so please start thinking about your questions to these fantastic people here. Mario you are also the regulator so and regulators can use sticks but they also can use carrots. So I guess my question for you is about the carrots.

Mario Nobile: Thank you. I’ll try to connect some dots and And my answer, I have three answers for this, but I want to connect. I fully agree with the other panelists, with Adam. The first one is education, and I connect with the answer from Adam. We are writing with public consultation guidelines for AI adoption, AI procurement, and AI development. So think about public administration. In Italy, we have 23,000 public administrations. Everyone must adopt. Everyone must do procurement. Some of them must develop. I think ministries like Environment and Energy, of course, but also the National Institute for Welfare. And what Adam was saying before is also about the interaction between cloud and edge computing. This is important. But we have also other challenges, and I’m thinking about the energy model. About which are the challenges now for a good use of artificial intelligence? Data quality and breaking the silos. These two points are really important for us. So the first answer is education, collaboration, sharing. New questions about what can I do with my data and with my data quality? What can I do about the cloud and the edge services I can develop? The second one, and the Agency for Digital Italy is working on it, is the Open Innovation Framework. It’s about new applications, not the classic tender about buying something, but the Open Innovation Framework, so I can enable public administration to carry for planned procedures and facilitate the supply and the demand in a new way. The third one is money. So the carrot is money. We are using in Italy a big amount of money from the National Recovery and Resilience Plan. We have 69 billion euros for ecological transition and 13 billion euros for business digitalization. Now we are working on a new framework about tax credit for enterprises for a good start of using AI in the small and medium enterprises. Thank you very much. This is fascinating, and also finishing with the money always gives people hope.

Moderator: But I wanted to underline the aspect of procurement, because in 99% of the UN member states, the public sector is still the biggest single buyer. So if we can have decent procurement policies, it’s already a lot, right? So congrats on that. Marco, on your side of the story, with your experience, if you can tell us what are the key enablers for this story, right? So what are the drivers? If someone needs to start looking into that, the policy makers and the scientists, what would you…

Marco Zennaro: and I’m going to be very concrete and practical. So the first one is investing in local capacity building in embedded AI. So capacity building has been mentioned many, many times in the last few days. But I would say the new aspect from my side is to give priority to funding for curricula that is co-developed with global partners. So not reinventing the wheel, but using the data to make the decision. So not reinventing the wheel, but using existing and open curricula such as the one that we developed with the TinyML Academic Network, which is completely open and can be reused. The second one is to promote open, low-cost infrastructure for TinyML deployment. So people need to have these devices in their hands, and very often that’s not easy. So, you know, subsidizing the access to these open source TinyML tools and low-power hardware would be extremely useful. Because, of course, this lowers the barriers, the entry barriers, and stimulates local innovation with these, you know, SDG-related challenges. The third one is to integrate TinyML into national digital and innovation strategies. So people, when they design a strategy, please don’t forget that you also have this kind of alternative model of really tiny devices running AI. So that’s, you know, a component that should be included. Next one is to fund context-aware pilot projects in key development sectors. We heard from Mozilla about, you know, funding kind of pilots and, you know, testing new solutions. Well, that’s possible for TinyML. Again, we have seen many, many interesting applications. So, you know, funding even more would be extremely useful. And finally, facilitating regional collaboration and knowledge sharing. So we had a few activities which were, like, tailored for specific regions with the idea that, you know, people from the same region have the same issues. And that has been extremely successful. So, you know, supporting this south-to-south collaboration and, you know, focusing on specific regions so that they can use TinyML. to solve their common issues.

Moderator: Super, very interesting. So co-working in many levels, right, as you said, but also interesting because I’ve been participating in several discussions about the DPI, digital public infrastructure, or also public interest infrastructures. And to be honest, I have not been hearing a lot about what you are saying. So I think it’s an interesting way also to connect the dots if this can be more and more presented as a potential solution. So Adam, let me get back to you now. I know that from where you are sitting in the Boston Consulting Group, you are also looking to governance models for this issue. So what are key features there of obviously in three minutes? It’s always a challenge, but that you can tell us on the governance side of the story.

Adham Abouzied: Yeah, yeah, very clear. I think we’ve been mainly looking at basically how to inspire what is the right governance for a systems level change to basically push forward, accelerate the adoption of open source data sharing, intellectual wealth sharing across different systems. And I think the most important thing and is to have the, I would say, the right policies and the right sharing protocols across every industry. And to be very well designed, well enforced, as you said before, with the carrot and the stick at the same time. It is also very important to make sure that the different players are actually incentivized to adopt. And actually, as Mario was saying also earlier, have the right, I would say, set up the right regulatory reference for them to adopt on their own level. And then afterwards, also sharing the outcomes, the insights, the data with others. We have faced so many difficulties in so many sectors while trying to apply AI applications, whether it’s generative or whether it’s other techniques with facing the current regulation. The cloud is one, basically having the proper data governance, data classification, understanding what is sensitive, what should be on the cloud, what should not be on the cloud and under which levels or layers of security. Even worse in several countries and specifically emerging countries that basically do not have hyperscalers or do not have actual physical wealth of data centers implemented locally and to have actual regulation against the data traveling, their data traveling cross border, wouldn’t even allow the, I would say, the prompts, the queries that would go up to query foundational models or other models that are sitting in the cloud somewhere outside of the country. So regulating this, what type of prompts should actually cross the border, which ones shouldn’t, and for which you maybe need to go for local alternatives that are more focused and smaller models, maybe less efficient. you start there until the regulation evolves is something that is that is very, very important. But I mean, after all, what is really, really important is to have certain rules across the players in a value chain about what they can share in which format. And what are the, I would say the rights and responsibilities that go with it as it moves through the value chain, and for them to believe that actually, sharing is a win-win situation. It is not the opposite of, I would say, a free market. I would say competitiveness and keeping your intellectual property and conservative, conserving your competitiveness, it actually contributes to it, it gives you a wealth of information on top of which a layer on top of which you can develop a competitive edge.

Moderator: Thank you. Very, very interesting. And again, I guess, Marco was with me yesterday in another panel about AI and the issue of standardization and rules appeared a lot, right? So we will need to deal with that rather sooner than later. So, Leona, again, as you noticed, the second round is a lot about lessons learned and insights for the different stakeholders. So in your case, what is what you would tell to policymakers from the perspective of what you are doing in Raido? What are key lessons to share with them?

Ioanna Ntinou: As you said, I have more of a technical background, right? So I think that one of the key lessons that I have learned from Raido or working as an AI developer is that sustainable AI is not going to emerge by default. We need to incentivize and it needs to be actively supported. And the reason that even when I train a model, I will opt for a bigger model. I will opt for more data. is the way we are actually measuring success, which is measured by accuracy. So if we are measuring everything by accuracy, and sometimes we neglect the cost that comes with accuracy, we might consume way more energy than what is actually needed. Like, in some cases, a small improvement comes with a huge increase in energy, and we have to be careful with this trade-off. And I think that one of the first steps that we need to do is to actually simply assess how much energy is used in widely used public models. And what I mean is that we… I will give you a concrete example, right? There is now GPT-4, or there is DeepSeq, that they might have 70 billion parameters, and they consume massive amounts of electricity during training and during inference, but we don’t really know how much energy is used when we put a simple prompt on GPT. We are not aware of this. So I think that we should start by having some transparency in the way we report energy. So by simply developing evaluation standards at prompt level, it will be a great first step, because it will develop awareness and transparency. And after this, we can see what can be done with this. But without having the knowledge of how much energy is actually used, I think we cannot focus on legislation, right?

Moderator: Super. Of course, I mean, it’s very important, because the transparency generates the evidence that is needed, right? And it’s a good segue for you, Marc, because the question for you was precisely how the civil society can be even stronger in demanding more transparency and accountability for this company.

Mark Gachara: Yeah, so I will answer the question in two parts. I will start with… The things that as Mozilla we see are missing so right now there’s there’s a lot of money going into climate or smart agriculture or climate AI solutions But it’s to make them more efficient more effective as opposed to how do we mitigate? The harms that they are causing and and the owner has just mentioned a little bit about that So It is important from our end to to think about how the solutions We build can prevent they can also improve transparency and then they can make Also it more visible over the impact on the environment over over the work that you are doing And this could come Through policy work through research and also strengthening community work like strengthening community civil society Organizations that are working with communities on the ground and I’ll give an example of somebody that we worked with Last year, but this is ongoing. So so in Kilifi Kenya, we have the Center for Justice governance and environmental action They Are working on climate justice issues and these plants in Kenya to build a nuclear reactor somewhere in Kilifi in the sea and and one of the things that this CSO did was to Do an ecological mapping of an area in the Indian Ocean and they used AI to do this and they have come up with a report and they are saying that We think that this is ill-advised and they’ve given the reasons why it’s ill-advised We should actually be focusing on using renewable energy sources because Kenya has is is net is a net producer of renewable energy, so Thank you very much. This is what civil society can do, for example. They are pushing back. It’s still a work in progress right now. But you can generate evidence to be able to do advocacy. Unfortunately, once the ship has left the dock for civil society, you are left trying to push back, which is unfortunate. But again, I come back to the three things. How do we use such kind of evidence to advise policy? How can we do action research? And how do we put money into preventive, into making these issues more transparent? So that actually the taxpayers and government can actually quantify the issue in front of them. Thank you.

Moderator: Thank you. Very interesting. And also the need to do that in an unfortunate, shrinking space for the civil society. And with less funds for this accountability and transparency, including for journalism. So now is the time for you. Questions? Leona, also on the online space, do we have questions for the panel? You have two mics here. I know that this setting looks like we are very distant, but we welcome your thoughts and questions. And Leona, from the online, I don’t know if you are hearing us.

Leona Verdadero: Yes, hi. No questions yet.

Moderator: So while maybe you guys are getting less shy, can I ask you, Leona, to be on the screen and give us a teaser of one minute about the report that we are going to launch soon about these issues?

Leona Verdadero: Yes. Hi. Good morning, everyone. Can you hear me? Yeah. Yes. Okay, great. Great to see all of the panelists. and thank you so much for joining. And I think just to also echo UNESCO’s work and really our work on being a laboratory of ideas and also trying to push the envelope on actually being able to define what do we mean by all these sustainable, low-resource, energy-efficient solutions. So yes, we’ve been doing this research in partnership with the University College of London where we’re doing experiments actually looking at how can we optimize the inference space, right, of AI. So inference is what other colleagues have mentioned since when we are interacting with these systems. But really what we are uncovering is which and I really love this conversation so far because most of you have echoed the need to be more efficient climate conscious and looking at AI in a smaller, stronger, smarter way. So here we’re looking at energy efficient techniques such as, you know, optimizing the models, making them smaller with quantization, distillation, all of these technical things. But what we are really seeing here is that these different experiments that we’ve done actually make the models smaller and also just more efficient and better performing. And what that means especially for stakeholders working in low resource settings, right, where you have limited access to compute and infrastructure, these types of models are made more accessible to you. And so it’s really also part of the answering the question, you know, what type of model, what type of AI do we need for the right type of job, right? So also trying to demystify the thought that, you know, bigger is better. It’s not necessarily better, right? So also that’s what you’re seeing now in terms of, you know, all these large language models are power hungry and water thirsty. So, you know, here’s where we’re trying to, you know, merge two things together. So one is very technical research, how we’re able to translate that in a policy setting, what it means for policymakers, if you’re thinking about exactly developing, deploying, procuring AI systems for your use cases. So if we’re able to, by presenting evidence, try to move the needle by promoting more eco-conscious AI choices, right, and model usage, then I think that’s one very concrete start to actually have this bigger push to look very concretely on what it means to use smaller and smarter AI. And so actually also in the chat, we’re going to put here a sign-up link so all of you will be notified when we launch this report. Thank you.

Moderator: Thank you, Leona, for the teaser, and I strongly recommend you’ll see the report. By the way, it’s also beautifully designed by one of our colleagues that is an expert in designing. So it shows, tried to show also through cartoons, etc., in a user-friendly way, some of these issues that are sometimes complex to understand. Questions from here? Otherwise, I have Brent, Jan. Please take a mic.

Audience: Hello, my name is Jan Lublinski. I work for DW Academy in Media Development under the roof of Deutsche Welle, but I’m also a former, in my earlier life, I was a science journalist, that’s 10 years ago, and I’m used to hearing new things, and I must say I’m thinking while I talk because most of what you told us is new to me. So that’s really exciting, and congratulations for the report and putting together the panel the way you have done, and I’m trying to remember other technological revolutions we’ve seen in the past and how they could be made more transparent, like, I don’t know, heating systems in houses. You know, in Germany we have passes for every house where you can see how energy-efficient a house is, and it’s obligatory in the EU, I think, and I think we can think of many other technologies where we made, step by step, made it more transparent to then actually see what it means so people take conscious decisions, and that’s why I’m really glad, again, that you… Stress the transparency issue, but of course next thought I have is we also trade carbon as a negative good, right? So we force the industry To become more efficient by giving them certificates on their carbon emissions, and I wonder would By everything you know now so far about this topic Do you think we can take transparency on energy consumption by this far that we can one? One day actually trade the right to consume energy on behalf of with AI technologies And then and by that way forced industry to you know develop more small models and and be really conscious of whether as you say We really need to take the last step that will be very expensive to get the very Extreme accuracy that we might not need so not in the way to advance in the direction that you all point to Maybe our colleague from Boston starting group has ideas on this as he’s looking at the overarching strategies. Thank you

Moderator: Thank you. Yeah, so we have five minutes and 39 seconds So it’s one minute per person to interact with Jan’s question and our maybe already Offer us one takeaway and maybe if you don’t necessarily want to interact with young questions I would ask you something related to what he said What are the questions you think science journalists should be asking about this issue So Mari over to you

Mario Nobile: Well in one minute is very tricky, but I’ll try I Think that the key words are related so we have sustainability calls for transparency calls for awareness calls for Policies also government policies, and I think that the good question is not when but now we must think about Sustainability, energy consumption, and the AI potential, the potential for AI to displace jobs. Everything is related. We cannot think about energy consumption without job losses. Okay, so I think that the journalist can ask for a solution to job losses, energy consumption, awareness from people. It’s very complicated. We need two hours and not one minute.

Moderator: Thank you, but interesting. And well, that’s the job of the journalists to find the time to do it.

Marco Zennaro: Well, my kind of question would be what kind of models do we need? So in TinyML, of course, in these small devices, you can only have like models which are very specific to an application. Again, you know, coffee leaves or, you know, turtles. So my point is, do we always need these super wide models that can answer every question we have? Or is it better to focus on models that solve specific issues which are useful for, you know, SDGs or for humanity in general?

Moderator: Small is beautiful, right? There was an article in the New Yorker a few years ago. Fantastic. I recommend not about this, about overall small is beautiful. Adam, your one minute.

Adham Abouzied: Yeah, I think it’s very interesting. I would reiterate again what I’m saying. I think, yes, smaller focused models is something that would create significant value and would be much more optimized to get there. They need to have access to data that is currently not necessarily out there. And for this to happen, we need to have in place the policy, the governance. that would make this happen, because there isn’t a wealth of data out there that would allow this. And honestly, to be, I mean, to consume the energy that is required for AI to deliver real value within the systems and value chains that helps develop on the SDGs and help the daily lives is a much more important, I would say, use than consuming this energy for the large models to create videos and images on the internet.

Moderator: Thank you. Ioana?

Ioanna Ntinou: I think that my question will be, as a researcher, if we focus so much on having smaller models, if we actually neglect all the progress that has been done so far with the large language models, what all the revolution that has done so far. But then I guess the answer to this is that task-specific small models have still a great value. And I’m actually, I’m talking about actual value. You see our mobile phones, the technology there is small models, because they are bounded from the battery life and the processor that the phone has. We can see that our phones are actually having small task-specific models, meaning that there is still a lot of things to learn, except for the energy part, in terms of science and knowledge and the things that we can achieve as humans from pushing the boundaries of research. So I also think that we should give another type of value on this, which is actually the knowledge and the learning out of this process.

Moderator: Thank you. Mark?

Mark Gachara: Yeah, on my end, to respond a bitto the question that was asked. The theater of where the most impact of climate is is in the global south and it would be a farmer, it would be indigenous and local communities. So it come back and say like how do we have funds that support grassroots organizers and indigenous communities to actually build some of these solutions together. It’s already been said like the problems are really localized and we can focus on local solutions and then we need to foster funding strategies that could actually look into research and creating science that actually solves this climate solutions with these local communities. And procurement has already been mentioned which was a good thing, public procurement. Thank you. Thank you so much.

Moderator: We come to the end of this fascinating discussion. I want to personally thank you a lot each one of you because I have learned a lot. I hope it was the same with the audience and those online. Thank you Adam, thank you Leona for your online participation and here Mario and Marco and Iona and Mark and for all of you to be listening to us attentively and let’s try to make this greener world while still trying to have the good benefits of this fantastic digital and AI revolution. Thank you so much. Enjoy the rest of the IGF. Thank you. Export animation Critical Shirt Critical Birth Critical Birth Main animation Critical Birth Last one

Agreement points

Need for smaller, more efficient AI models over large general-purpose models

Speakers

– Mario Nobile
– Marco Zennaro
– Adham Abouzied
– Ioanna Ntinou
– Leona Verdadero

Arguments

Transition from large energy-consuming models to vertical and agile foundation models for specific purposes like health, transportation, and manufacturing

Question whether super-wide models answering every question are needed versus specific models solving targeted issues

AI optimization in energy sector from generation to transmission to usage requires cooperation and data exchange across value chains

Knowledge distillation can reduce model parameters by 60% while maintaining accuracy, as demonstrated in energy grid forecasting

UNESCO’s upcoming report “Smarter, Smaller, Stronger” demonstrates that optimized models can be smaller, more efficient, and better performing

Summary

All speakers agreed that the future of sustainable AI lies in developing smaller, task-specific models rather than continuing to scale up large general-purpose models. They emphasized that these smaller models can maintain effectiveness while dramatically reducing energy consumption.

Topics

Development | Infrastructure | Economic

Critical importance of transparency and measurement in AI energy consumption

Speakers

– Ioanna Ntinou
– Mark Gachara
– Moderator

Arguments

Need for transparency in energy reporting and evaluation standards at prompt level to develop awareness of AI energy usage

Developers can optimize code energy consumption through tools like Code Carbon to measure environmental impact

Transparency in technology adoption generates evidence needed for effective policymaking, similar to energy efficiency standards in other sectors

Summary

Speakers unanimously agreed that without proper measurement and transparency of AI energy consumption, it’s impossible to make informed decisions about AI usage or develop appropriate legislation. They emphasized the need for standardized measurement tools and reporting mechanisms.

Topics

Legal and regulatory | Development | Infrastructure

Necessity of open source and collaborative approaches for sustainable AI development

Speakers

– Marco Zennaro
– Adham Abouzied
– Mark Gachara

Arguments

Investment in local capacity building using open curricula co-developed with global partners prevents reinventing the wheel

Open source solutions optimize energy costs by avoiding repetitive development, potentially saving trillions in licensing costs

Developers can optimize code energy consumption through tools like Code Carbon to measure environmental impact

Summary

Speakers agreed that open source approaches are essential for sustainable AI development, as they prevent duplication of effort, reduce costs, and enable collaborative problem-solving while minimizing energy consumption through shared resources and knowledge.

Topics

Economic | Development | Infrastructure

Need for comprehensive policy frameworks and governance structures

Speakers

– Mario Nobile
– Adham Abouzied
– Marco Zennaro

Arguments

Need for guidelines on AI adoption, procurement, and development across 23,000 public administrations in Italy

Right policies and sharing protocols across industries are needed with proper regulatory frameworks and incentives

Integration of TinyML into national digital and innovation strategies is essential for comprehensive AI planning

Summary

All speakers emphasized the critical need for comprehensive governance frameworks that include proper policies, guidelines, and regulatory structures to support sustainable AI development and deployment across different sectors and scales.

Topics

Legal and regulatory | Economic | Development

Similar viewpoints

Both speakers emphasized the importance of supporting local and regional communities, particularly in the Global South, to develop context-appropriate AI solutions that address their specific challenges and needs.

Speakers

– Marco Zennaro
– Mark Gachara

Arguments

Regional collaboration and south-to-south knowledge sharing has proven extremely successful for TinyML applications

Support for grassroots organizers and indigenous communities to build localized climate solutions through targeted funding

Topics

Development | Human rights | Sustainable development

Both speakers advocated for innovative approaches to cooperation and procurement that move beyond traditional methods to enable better collaboration and data sharing across organizations and sectors.

Speakers

– Mario Nobile
– Adham Abouzied

Arguments

Open Innovation Framework enables new procurement approaches beyond traditional tenders for public administration

System-level cooperation and data sharing across value chains is essential for meaningful vertical AI models

Topics

Legal and regulatory | Economic | Development

Both speakers emphasized that sustainable AI development requires deliberate intervention and active promotion, challenging the current paradigm that prioritizes model size and accuracy over efficiency and sustainability.

Speakers

– Ioanna Ntinou
– Leona Verdadero

Arguments

Sustainable AI will not emerge by default and needs active support and incentivization beyond just accuracy metrics

UNESCO’s upcoming report “Smarter, Smaller, Stronger” demonstrates that optimized models can be smaller, more efficient, and better performing

Topics

Development | Legal and regulatory | Economic

Unexpected consensus

Integration of social and environmental concerns in AI development

Speakers

– Mario Nobile
– Mark Gachara
– Ioanna Ntinou

Arguments

Need to address job displacement alongside energy consumption as interconnected challenges requiring comprehensive solutions

Civil society can use AI for ecological mapping and evidence generation to advocate against environmentally harmful projects

Focus on smaller, task-specific models while not neglecting progress made with large language models

Explanation

Unexpectedly, speakers from different backgrounds (government regulator, civil society advocate, and technical researcher) all emphasized the need to consider AI’s social, environmental, and technical impacts as interconnected rather than separate issues. This holistic view was surprising given their different professional perspectives.

Topics

Economic | Development | Sociocultural

Practical implementation focus over theoretical discussions

Speakers

– Mario Nobile
– Marco Zennaro
– Ioanna Ntinou
– Mark Gachara

Arguments

Italy allocates 69 billion euros for ecological transition and 13 billion for business digitalization from National Recovery Plan

TinyML applications include disease detection in livestock, bee counting, anemia detection, and turtle behavior monitoring across 60+ universities in 32 countries

Smaller models deployed in smart homes for energy demand forecasting help companies make more accurate production decisions

Civil society can use AI for ecological mapping and evidence generation to advocate against environmentally harmful projects

Explanation

All speakers, regardless of their institutional background, focused heavily on concrete, practical applications and real-world implementations rather than theoretical discussions. This consensus on pragmatic approaches was unexpected in an academic/policy setting where abstract discussions often dominate.

Topics

Development | Infrastructure | Economic

Overall assessment

Summary

The speakers demonstrated remarkable consensus across multiple key areas: the need for smaller, more efficient AI models; the critical importance of transparency and measurement; the value of open source and collaborative approaches; and the necessity of comprehensive policy frameworks. They also shared unexpected agreement on integrating social and environmental concerns and focusing on practical implementation.

Consensus level

High level of consensus with strong implications for sustainable AI development. The agreement across speakers from different sectors (government, academia, civil society, private sector) suggests these principles have broad support and could form the foundation for coordinated action on sustainable AI. The consensus indicates a mature understanding of the challenges and a convergence toward practical, implementable solutions rather than competing approaches.

Different viewpoints

Balance between large foundational models and small specialized models

Speakers

– Ioanna Ntinou
– Marco Zennaro
– Adham Abouzied

Arguments

Focus on smaller, task-specific models while not neglecting progress made with large language models

Question whether super-wide models answering every question are needed versus specific models solving targeted issues

AI breakthroughs came from training on wealth of internet data; for vertical impact, models need to see and train on data across value chains

Summary

Ioanna expressed concern about neglecting progress from large language models while focusing on smaller ones, whereas Marco questioned the necessity of super-wide models, and Adham emphasized the value of large foundational models for breakthrough innovations while supporting vertical specialization

Topics

Development | Infrastructure | Economic

Unexpected differences

Scope of open source collaboration versus national strategic control

Speakers

– Mario Nobile
– Adham Abouzied
– Marco Zennaro

Arguments

Need for guidelines on AI adoption, procurement, and development across 23,000 public administrations in Italy

Open source solutions optimize energy costs by avoiding repetitive development, potentially saving trillions in licensing costs

Investment in local capacity building using open curricula co-developed with global partners prevents reinventing the wheel

Explanation

While all speakers supported collaboration, there was an unexpected tension between Mario’s emphasis on national strategic control and standardized guidelines versus Adham and Marco’s push for more open, collaborative approaches that transcend national boundaries

Topics

Legal and regulatory | Economic | Development

Overall assessment

Summary

The discussion showed remarkable consensus on core goals (energy efficiency, sustainability, accessibility) with disagreements primarily focused on implementation approaches and the balance between different technical strategies

Disagreement level

Low to moderate disagreement level. Most conflicts were constructive and focused on technical approaches rather than fundamental goals. The main tension was between preserving innovation from large models while promoting efficiency through smaller ones. This suggests a healthy debate that could lead to complementary rather than competing solutions, with implications for developing comprehensive AI sustainability policies that accommodate multiple technical approaches.

Partial agreements

Similar viewpoints

Both speakers emphasized the importance of supporting local and regional communities, particularly in the Global South, to develop context-appropriate AI solutions that address their specific challenges and needs.

Speakers

– Marco Zennaro
– Mark Gachara

Arguments

Regional collaboration and south-to-south knowledge sharing has proven extremely successful for TinyML applications

Support for grassroots organizers and indigenous communities to build localized climate solutions through targeted funding

Topics

Development | Human rights | Sustainable development

Both speakers advocated for innovative approaches to cooperation and procurement that move beyond traditional methods to enable better collaboration and data sharing across organizations and sectors.

Speakers

– Mario Nobile
– Adham Abouzied

Arguments

Open Innovation Framework enables new procurement approaches beyond traditional tenders for public administration

System-level cooperation and data sharing across value chains is essential for meaningful vertical AI models

Topics

Legal and regulatory | Economic | Development

Both speakers emphasized that sustainable AI development requires deliberate intervention and active promotion, challenging the current paradigm that prioritizes model size and accuracy over efficiency and sustainability.

Speakers

– Ioanna Ntinou
– Leona Verdadero

Arguments

Sustainable AI will not emerge by default and needs active support and incentivization beyond just accuracy metrics

UNESCO’s upcoming report “Smarter, Smaller, Stronger” demonstrates that optimized models can be smaller, more efficient, and better performing

Topics

Development | Legal and regulatory | Economic

Key takeaways

AI governance requires a multi-pillar approach combining education, research, public administration, and enterprise engagement to ensure inclusive growth

Energy-efficient AI solutions like TinyML and model optimization can significantly reduce power consumption while maintaining performance – demonstrated by 60% parameter reduction with maintained accuracy

Open source approaches can save trillions in development costs by avoiding repetitive creation of AI solutions across organizations

Transparency in AI energy consumption reporting is essential before effective legislation can be developed – current lack of visibility into energy usage per prompt or model interaction

Smaller, task-specific AI models often provide better value than large general-purpose models for solving specific problems like healthcare, agriculture, and environmental monitoring

System-level cooperation and data sharing across value chains is necessary for vertical AI models to create meaningful impact in sectors like energy optimization

Civil society can leverage AI tools for evidence generation and advocacy, particularly for environmental justice issues in developing regions

Public procurement policies represent a powerful lever for promoting sustainable AI adoption given governments’ role as largest buyers

Resolutions and action items

UNESCO and University College London to launch ‘Smarter, Smaller, Stronger’ report on resource-efficient generative AI

Italy to implement guidelines for AI adoption, procurement, and development across 23,000 public administrations

Italy developing tax credit framework for small and medium enterprises to incentivize AI adoption

Promotion of open Innovation Framework to facilitate new procurement approaches beyond traditional tenders

Integration of TinyML into national digital and innovation strategies

Investment in local capacity building using open curricula co-developed with global partners

Funding for context-aware pilot projects in key development sectors

Development of evaluation standards for energy consumption at prompt level

Unresolved issues

How to balance the trade-off between AI model accuracy and energy consumption when accuracy remains the primary success metric

Regulatory challenges around data governance, cloud usage, and cross-border data transfer that limit AI implementation

Lack of transparency in energy consumption reporting for widely used public AI models like GPT-4

How to address job displacement concerns alongside energy consumption as interconnected challenges

Whether focusing on smaller models might neglect important progress made with large language models

How to ensure adequate funding reaches grassroots organizations and indigenous communities for localized climate solutions

Implementation of carbon trading mechanisms for AI energy consumption similar to other industries

Suggested compromises

Transition gradually from large energy-consuming models to smaller vertical models while maintaining research progress on both fronts

Use large models as ‘teachers’ through knowledge distillation to create smaller ‘student’ models that maintain performance

Implement hybrid approaches combining cloud and edge computing to optimize energy usage

Focus AI energy consumption on delivering real value within systems that support SDGs rather than general content generation

Develop task-specific models for mobile and resource-constrained environments while continuing research on larger models

Balance accuracy requirements with energy costs by questioning whether maximum accuracy is always necessary for specific use cases

Now the debate is not humans versus machines. Now the debate is about who understands and uses managed AI versus machines. who don’t.

Speaker

Mario Nobile

Reason

This reframes the entire AI discourse from a fear-based narrative about AI replacing humans to a more nuanced understanding about digital literacy and AI competency. It shifts focus from existential concerns to practical skill development and education.

Impact

This comment established education as a central theme throughout the discussion. It influenced subsequent speakers to emphasize capacity building, local training, and the democratization of AI knowledge, moving the conversation from technical solutions to human empowerment.

The study basically estimates that if we would recreate only one time the wealth of open source intellectual property that is available today on the Internet, it would cost us $4 billion. But it does not stop there… you would increase this $4 billion of cost to $8 trillion just because of the repetition.

Speaker

Adham Abouzied

Reason

This provides concrete economic evidence for the value of open-source collaboration, transforming an abstract concept into tangible financial terms. It demonstrates how sharing and collaboration can create exponential value while reducing resource consumption.

Impact

This comment shifted the discussion from individual technical solutions to systemic collaboration. It influenced other panelists to emphasize knowledge sharing, regional cooperation, and the importance of breaking down silos between organizations and sectors.

Sustainable AI is not going to emerge by default. We need to incentivize and it needs to be actively supported… if we are measuring everything by accuracy, and sometimes we neglect the cost that comes with accuracy, we might consume way more energy than what is actually needed.

Speaker

Ioanna Ntinou

Reason

This challenges the fundamental assumption in AI development that bigger and more accurate is always better. It exposes the hidden costs of the current success metrics and calls for a paradigm shift in how we evaluate AI systems.

Impact

This comment introduced critical thinking about measurement frameworks and success criteria. It led to discussions about transparency in energy reporting and the need for new evaluation standards that balance performance with sustainability.

Do we always need these super wide models that can answer every question we have? Or is it better to focus on models that solve specific issues which are useful for, you know, SDGs or for humanity in general?

Speaker

Marco Zennaro

Reason

This question fundamentally challenges the current trajectory of AI development toward ever-larger general-purpose models. It connects technical decisions to broader humanitarian goals and sustainable development.

Impact

This question became a recurring theme that influenced multiple speakers to advocate for task-specific, smaller models. It helped establish the ‘small is beautiful’ philosophy that ran through the latter part of the discussion and connected technical choices to social impact.

The theater of where the most impact of climate is is in the global south and it would be a farmer, it would be indigenous and local communities… we need to foster funding strategies that could actually look into research and creating science that actually solves this climate solutions with these local communities.

Speaker

Mark Gachara

Reason

This comment brings crucial equity and justice perspectives to the technical discussion, highlighting that those most affected by climate change are often least represented in AI development. It challenges the panel to consider who benefits from and who bears the costs of AI solutions.

Impact

This intervention grounded the entire discussion in real-world impact and social justice. It influenced the conversation to consider not just technical efficiency but also accessibility, local relevance, and community empowerment in AI solutions.

Overall assessment

These key comments fundamentally shaped the discussion by challenging conventional assumptions about AI development and success metrics. They moved the conversation from a purely technical focus to a more holistic view that encompasses education, collaboration, sustainability, and social justice. The comments created a progression from individual technical solutions to systemic thinking about governance, measurement, and community impact. Most significantly, they established a counter-narrative to the ‘bigger is better’ approach in AI, advocating instead for targeted, efficient, and socially conscious AI development that prioritizes real-world problem-solving over technical prowess.

How can we transition from brute force models to vertical and agile foundation models with specific purposes across different sectors like health, transportation, tourism, and manufacturing?

Speaker

Mario Nobile

Explanation

This is crucial for reducing energy consumption while maintaining AI effectiveness in specific domains, representing a key strategic shift in AI development approach.

How can we effectively break data silos and improve data quality to enable better AI applications in public administration?

Speaker

Mario Nobile

Explanation

Data quality and interoperability are fundamental challenges that need to be addressed for successful AI implementation in government services.

What are the specific energy consumption metrics at the prompt level for widely used public models like GPT-4?

Speaker

Ioanna Ntinou

Explanation

Without transparency in energy reporting at the individual query level, it’s impossible to make informed decisions about sustainable AI usage and develop appropriate legislation.

Can we develop a carbon trading system specifically for AI energy consumption to incentivize more efficient models?

Speaker

Jan Lublinski (Audience)

Explanation

This would create market-based incentives for developing energy-efficient AI systems, similar to existing carbon trading mechanisms in other industries.

Do we always need super wide models that can answer every question, or is it better to focus on models that solve specific issues useful for SDGs?

Speaker

Marco Zennaro

Explanation

This fundamental question challenges the current trend toward large general-purpose models and could reshape AI development priorities toward more sustainable, task-specific solutions.

How can we establish proper data governance and classification standards for cross-border AI applications, especially in emerging countries?

Speaker

Adham Abouzied

Explanation

Current regulations often prevent effective AI implementation due to unclear data governance rules, particularly affecting countries without local hyperscaler infrastructure.

What evaluation standards should be developed to assess energy consumption at the prompt level for AI models?

Speaker

Ioanna Ntinou

Explanation

Standardized evaluation methods are needed to create transparency and enable comparison of energy efficiency across different AI systems.

How can funding strategies be developed to support grassroots organizers and indigenous communities in building localized AI climate solutions?

Speaker

Mark Gachara

Explanation

Since climate impacts are most severe in the Global South and among indigenous communities, research funding should prioritize supporting these communities in developing their own solutions.

What are the optimal policies and sharing protocols needed across industries to accelerate adoption of open source AI solutions?

Speaker

Adham Abouzied

Explanation

System-level cooperation requires well-designed governance frameworks to incentivize data and intellectual property sharing while maintaining competitive advantages.

How can we balance the trade-off between model accuracy and energy consumption in AI development?

Speaker

Ioanna Ntinou

Explanation

Current success metrics focus primarily on accuracy, often neglecting the disproportionate energy costs of marginal improvements, requiring new evaluation frameworks.

Summary

This lightning talk focused on Universal Acceptance and Internationalized Domain Names (IDNs), presented by Regina Filipova Fuchsova from the .eu registry and Esteve Sanz from the European Commission. The discussion centered on the importance of enabling internet users worldwide to navigate online entirely in their local languages, moving beyond ASCII-only domain names to support various scripts including Cyrillic, Greek, Chinese, and Arabic characters. Universal Acceptance refers to the principle that all domain names and email addresses should be treated equally by internet-enabled applications, devices, and systems, regardless of the script or characters used.

Sanz emphasized that multilingualism is fundamental to EU identity, with the bloc’s motto being “united in diversity” across 24 official languages. He highlighted that 88% of European country-code top-level domains (ccTLDs) technically support IDNs, yet adoption remains slow due to limited user awareness. The speakers presented findings from the freshly published IDN World Report 2025, which revealed concerning trends including flat or negative growth in IDN registrations, high market concentration among a few top-level domains, and persistent technical barriers.

The report found an estimated 4.4 million IDNs worldwide, with 70% under ccTLDs, but showed negative growth rates of nearly 1% for ccTLDs and 5.5% for generic TLDs. Technical challenges persist, with over half of registries not supporting Unicode email addresses and three-fourths not permitting Unicode symbols in contact emails. The main barriers to adoption were identified as technical compatibility issues, low public awareness, and entrenched habits of using ASCII characters. The presenters concluded with a call for action to share IDN stories and expand data collection to better understand and promote multilingual internet adoption.

Keypoints

**Major Discussion Points:**

– **Universal Acceptance and Internationalized Domain Names (IDNs)**: The core focus on enabling domain names and email addresses in non-ASCII scripts (like Cyrillic, Greek, Arabic, Chinese) to work seamlessly across all internet applications and systems, allowing users to navigate the internet entirely in their local languages.

– **European Union’s Multilingual Digital Policy**: Discussion of how the EU’s “united in diversity” motto translates into digital infrastructure, with 24 official languages requiring support, and how regulations like the Digital Services Act incorporate provisions for linguistic diversity online.

– **IDN World Report 2025 Findings**: Presentation of key research results showing concerning trends including negative growth rates for IDNs (-1% for ccTLDs, -5.5% for gTLDs), high market concentration (top 10 TLDs account for 75% of all IDNs), and significant technical barriers to adoption.

– **Technical and Adoption Barriers**: Identification of major obstacles including compatibility issues with browsers and email services, low public awareness, entrenched habits of using ASCII/English domains, and insufficient promotion by registries and registrars due to limited business cases.

– **Cultural Preservation and Digital Inclusion**: Emphasis on preserving linguistic heritage online, with examples like Sámi language and the concept that local language internet access could potentially reduce internet shutdowns by increasing user connection to digital spaces.

**Overall Purpose:**

The discussion aimed to raise awareness about Universal Acceptance and IDN implementation in Europe, present research findings on the current state of multilingual internet infrastructure, and advocate for better support of linguistic diversity online to ensure digital inclusion for all language communities.

**Overall Tone:**

The tone was professional and educational throughout, with speakers maintaining an informative and collaborative approach. The presenters demonstrated expertise while acknowledging challenges and limitations in their research. The tone remained constructive even when discussing concerning trends like negative growth rates, and concluded with an encouraging call-to-action for community involvement and cooperation.

Speakers

– **Regina Filipova Fuchsova**: Works for the .eu registry, involved in Universal Acceptance and IDN (Internationalized Domain Names) research and advocacy

– **Esteve Sanz**: Works for the European Commission, focuses on EU digital priorities and multilingual internet policies, from Catalonia region of Spain

– **Tapani Tarvainen**: From Electronic Frontier Finland, also a non-commercial stakeholder group member in ICANN, expertise in internet governance and minority language digital rights

Additional speakers:

None identified beyond the speakers names list.

# Universal Acceptance and Internationalized Domain Names Discussion

## Introduction and Context

This lightning talk session focused on Universal Acceptance and Internationalized Domain Names (IDNs), featuring presentations by Regina Filipova Fuchsova from the .eu registry and Esteve Sanz from the European Commission. The session centered on enabling internet users worldwide to navigate online in their local languages, moving beyond ASCII-only domain names to support diverse scripts including Cyrillic, Greek, Chinese, and Arabic characters. The discussion included questions from audience members, notably Tapani Tarvainen of Electronic Frontier Finland and ICANN’s non-commercial stakeholder group.

## Core Technical Concepts

### Universal Acceptance Framework

Regina Filipova Fuchsova explained that Universal Acceptance refers to the principle whereby all domain names and email addresses should be treated equally by internet-enabled applications, devices, and systems, regardless of the script or characters used. This framework enables domain names to utilize special characters and different scripts beyond ASCII. Unicode versions have to be transferred via so-called Punicode to ASCII characters for DNS processing.

However, significant technical barriers remain. Over half of registries do not support Unicode addresses in email servers, while three-fourths of registries do not permit Unicode symbols as contact emails in their registry database. These limitations represent fundamental obstacles to achieving Universal Acceptance.

## European Union’s Multilingual Digital Policy

### Policy Foundation

Esteve Sanz emphasized that multilingualism forms the cornerstone of EU identity, with the bloc’s motto “united in diversity” encompassing 24 official languages. The European Commission requires all official EU content to be available in every official language, demonstrating institutional commitment to digital linguistic equality.

Sanz noted that the Digital Services Act includes provisions that incentivize platforms to promote linguistic diversity. He also highlighted that most European domains technically support IDNs, with 88% of European ccTLDs able to register names with local characters according to the latest statistics.

## IDN World Report 2025: Key Findings

### Market Statistics and Trends

Regina Filipova Fuchsova presented findings from the IDN World Report 2025, a recurring research project produced by URID (the .eu registry) with partners including UNESCO since 2011. The report, which has evolved from PDF format to a website, revealed concerning trends in the global IDN landscape.

The report estimated 4.4 million IDNs worldwide, with 70% operating under country-code top-level domains (ccTLDs). However, the market shows negative growth rates, with almost minus 1% yearly growth for ccTLDs and minus 5.5% for generic TLDs.

The IDN market exhibits high concentration, with the top 10 TLDs accounting for 75% of all IDNs globally. For the .eu registry specifically, IDN adoption remains at approximately 1% of total registrations, despite technical support being available.

### Technical Implementation Status

The report identified that 67% of Whois services display both Unicode and Punicode versions of domain names. However, significant gaps remain in email infrastructure support, with many registries unable to handle Unicode addresses or contact emails.

## Barriers to IDN Adoption

Three primary barriers hinder IDN adoption:

– Technical compatibility issues with browsers and email services

– Low public awareness of IDN capabilities

– Entrenched habits favoring ASCII/English domains

Additionally, insufficient promotion by registries and registrars, often due to unclear business cases, contributes to low adoption rates.

## Cultural Preservation and Minority Languages

### Indigenous Language Concerns

The discussion emphasized multilingual internet access as crucial for digital inclusion and cultural heritage preservation. Tapani Tarvainen raised specific concerns about minority language support, referencing the Sámi song performed at Monday’s opening ceremony at the town hall and questioning ICANN’s policies regarding IDN top-level domains.

Tarvainen highlighted that small minority languages face particular challenges, noting that characters like the Sámi letter Eng may become less usable due to lack of advocacy, observing that “nobody out there cares” about very small minority languages.

Regina Filipova Fuchsova acknowledged that IDN variant tables, last updated in 2008, require updating to better support special characters used in minority languages.

### Digital Rights Perspective

Esteve Sanz shared an interesting perspective he had heard from others about the connection between local language internet access and internet shutdowns. He explained: “We’re very worried about Internet shutdowns… And very intelligent people that know the local characteristics of these shutdowns, they tell us that if the Internet was in our local language, we would not see a shutdown. Because people would really feel the Internet as themselves, and politicians could not perform that shutdown.”

## Call to Action

Regina Filipova Fuchsova made specific requests for expanding IDN research and awareness:

– More TLDs to participate in IDN World Report data collection

– Communities to share their IDN stories and case studies for publication on the IDN World Report website

– Help reaching out to more ccTLDs to participate in annual research surveys

She also committed to follow-up discussion on Sámi language character support in IDN policies, addressing the minority language concerns raised during the session.

## Conclusion

The discussion highlighted both the technical capabilities and practical challenges of implementing Universal Acceptance and IDN adoption. While strong policy support exists, particularly within the EU framework, significant barriers remain in technical implementation, commercial incentives, and user awareness. The specific concerns raised about minority language support demonstrate how technical governance decisions can impact linguistic diversity and cultural preservation online.

Moving forward, bridging the gap between policy intentions and practical implementation will require coordinated efforts across technical, commercial, and advocacy domains to ensure that Universal Acceptance principles benefit all language communities.

Regina Filipova Fuchsova: Good morning, ladies and gentlemen. Welcome to this lightning talk on Universal Acceptance and IDN Word Report. My name is Regina Fucsova, I work for the .eu registry and together with Esteve Sanz from the European Commission, we will try our best to put this topic in the European context and also provide you with some exclusive insights in the freshly published IDN Word Report 2025. Before we dive into the world of IDNs and Universal Acceptance, let me just shortly put us in the place where we speak about in the context of Internet infrastructure. You can see an example of a web address leading to also quite freshly published report on the digital decade by the European Commission. And if we start from the right hand side, you can see .eu, which we refer to as top level domain, and to the left hand side, Europa. This Europa is referred to as second level domain. Together, Europa.eu is what we commonly refer to as a domain name. And this is also something what you read as the ccTLD registries. The top level domain doesn’t need to have only two letters. We have .org, .net, or maybe you saw around a booth with, for example, .post, one of the new gTLDs. To have it a bit more complicated even, this domain name doesn’t need to be in ASCII codes only. It can have special characters, such as this O in Lilleström, or it can be in a completely different script, Chinese, Arabic, or if we speak about the official European Union languages, Cyrillic or Greek. Another example you can see on the slide is a favorite salty snack, soleti.eu in a Bulgarian language. This soleti.eu in Cyrillic is referred to as Unicode version of the domain name. But for the DNS system to process it, it has to be transferred via so-called Punicode to ASCII characters, and you can see that it starts with xn and two dashes. This will be important for some of the findings of the report later on. In any case, IDNs are necessary for multilingual Internet. The concept of universal acceptance goes a bit further. It says that all domain names and also email addresses should be treated equally and used by all Internet-enabled applications, devices, and systems, regardless of the script or characters. The point is that users around the world should be able to navigate the Internet entirely in their local language. It has a big relevance to IGF and the WSIS. For example, on Tuesday in one of the sessions, it was among the two most voted WSIS-related initiatives. Universal acceptance included Internet infrastructure as for the initiatives which need more support. It was considered as the way how to attract the next billion of Internet users, and throughout this week, you might have repeatedly heard in the sessions the words like linguistic accessibility, digital inclusion, part of a concept that nobody is left behind, both in Internet governance and usage of the Internet, and it has also a strong link to sustainable development goals. Do you remember this lady from our Monday opening ceremony at the town hall? Just to illustrate that multilingualism has a face, we enjoyed a very beautiful song sung in a Sámi language. It’s very important for Norway because most of the Sámi population lives in Norway. It’s actually the only indigenous peoples in the EU, and I bet that you would agree with me that even for enjoying songs in English, it’s a very nice experience. It would be a big pity not to be able to enjoy the one which we heard on Monday. And since the offline and online world are more and more interconnected, then it’s just a logical thing to try our best as much as the technical equipment allows us to preserve the cultural heritage also in the online world. Now I would like to give the floor to Esteve to tell us how this concept is in line with the European digital priorities.

Esteve Sanz: Thank you so much, Regina. Thank you so much for hosting this lightning talk. I’m really looking forward to hear more about the report, which I don’t know about yet, so that will be a premiere for me as well. The EU, DNA really, our identity is based on multilingualism. Our motto is united in diversity. We have 24 official languages in the EU. There are many more languages, and this multilingual priority of the EU carries, of course, into the digital realm. EU policy and funding strongly support online multilingualism in many different aspects, in universal acceptance, but also in other regulations that I will talk about in a minute. And we require, of course, that all official EU content in every language, official language of the EU, and the capacity for everyone intervening in the policy process to exercise that right in that language. This is the DNA, really, of the EU multilingualism, and we are very proud to also support universal acceptance as a core value. As Regina was explaining, universal acceptance means all domain names and email addresses work everywhere online, whether in Latin, Greek, Cyrillic, or any script. This is crucial for Europe, where scripts and accented characters vary, so that French accents, Greek or Cyrillic languages work as smoothly as example.com. Today, 68% of domain registries say universal acceptance is the top factor for boosting multilingual domains. We need every app, browser, and form to accept all EU languages. Most European domains technically support IDNs, 88% of European ccTLDs, according to the latest statistics, can register names with local characters, yet uptake is slow. User awareness of universal acceptance is normally what we are told is the limiting factor. that it’s blocking that, and that’s why we’re happy to have this lightning talk to address that awareness. There are interesting, successful stories in the multilingual internet space in the EU. I’m from Catalonia, a region of Spain, and Catalan has built an extremely strong online presence. There is a dedicated .cat domain with 100,000 sites, and today 66% of websites in Catalonia have a Catalan version, up to only 39% in 2002. We have successful examples in Irish Gaelic, which went from limited to use in the EU to a full EU language, and then this also boosts online presence. There is a lot of coordination with member states. Many countries are championing their language online, whether Spain is supporting co-official languages like Catalan and Basque, or Baltic countries developing AI for the language, for their languages. The EU complements this funding. For example, we have the European language data space, but also injects in the new rules that we have this multilingual element. It’s not well known, but the Digital Services Act, which is our flagship when it comes to digital regulation, has several provisions that incentivize platforms to actually promote linguistic diversity. So, you know, situations, the policies are there, the political intentions are there. We need to bridge that gap. This is extremely important for the EU. It’s also extremely important, if I have to say, for things that we normally don’t think about. So we’ve engaged a lot with Global South players these days in the IGF, as we constantly do. And we heard one very interesting idea that I just wanted to put to you. We’re very worried about Internet shutdowns, that, you know, despite all these commitments, they continue to be on the rise. And very intelligent people that know the local characteristics of these shutdowns, they tell us that if the Internet was in our local language, we would not see a shutdown. Because people would really feel the Internet as themselves, and politicians could not perform that shutdown. I thought that was a very powerful idea.

Regina Filipova Fuchsova: It’s a very interesting aspect. Thank you. Maybe we can consider it for the future to have a look in the study in this aspect. It actually hasn’t come to our mind. Thank you very much. Thank you very much, Esteve, for your explanation. So, in the remaining time, I would like to show you the findings of the report. Just briefly, why URID cares at all, and who we are. I mentioned that we are running .eu, a top-level domain. We are already since 2005 by the appointment of the European Commission, and from the very beginning, we are committed to provide our support in all official EU languages. Throughout the years, we have introduced first IDNs under .eu, then also the Cyrillic version of the top-level domain, and then also the Greek version. In total, we account for 3.7 million of registrations, and about 1% out of them are IDNs. This might seem really low, 1%, but it’s actually in line with what our CCTLDs are experiencing. Also taking into account the technical and other issues, the uptake is not as it could be. And it’s a part of our strategy to bring European values to open, safe, and inclusive Internet. So this corresponds together. The IDN World Report already has a history since 2011, when we started it at that time in a PDF printed form together with UNESCO and other partners. We have got more partners over the years. The aim is to enhance the linguistic diversity through advocacy of internationalized domain names and their usage. This report, which is now in the form of a website, you can see on the right-hand side idnworldreport.eu brings news, registration trends, also information on technical readiness, and also we have added quite recently a kind of sentiment analysis. The outcomes of the 2025 report can be on a very high level grouped in three main topics. One is that many TLDs are experiencing flat or even negative growth trends. Second, the IDN market is highly concentrated, so to say in the hands of a few TLDs. And third, there are still challenges in user awareness and adoption. Some illustration of this, and I have to say at the beginning that the study has its limitation by the number of TLDs which provide us with the data. So this is also a call for action. Whoever from you can address in your communities people and organizations dealing with IDNs, please direct us to our study. So based on the data set, we had over 400 TLDs. There is an estimated 4.4 million of IDNs worldwide, and roughly 70% are under country code TLDs. The growth is indeed negative, almost minus 1% for yearly growth for ccTLDs, and even 5.5% for gTLDs. The top 10 TLDs, which is in line with the concentration I mentioned, account for over 3.3 million of domain names, so 75% more or less of all the IDNs. And the top 5 IDNs are in the Cyrillic version of Russian top-level domain .com.de and the 4th and 5th place belongs to Chinese registrations. We are also following technical aspects of IDN implementation and found out that over half of the registries do not support Unicode addresses in Vail mail servers at all. Unicode was with Soleti.eu in the Cyrillic script, so the form which is readable for humans. Three-fourths of the registries do not permit Unicode symbols as contact emails in their registry database, and actually none of the ccTLDs stated that they would offer support to EIA, stands for email addresses in international form, so it means that they have this Unicode quotation also in the email addresses also before the ad. And the last piece of information in these slides refers to the display of IDNs in this form of Unicode and also the Unicode xn-quotation I was mentioning. When it comes to Whois services, when the registries display both versions in 67% of cases and slightly lower rates are for registry databases. interfaces display and user interfaces display. So only on this information you can already see why the uptake is not as it could be, because if you cannot use the domain name and the connected email address in all the services you are used to, not only connected to registration, but also to the services by big platforms, for example, then it makes less sense. And then also the users who otherwise cannot access internet because of the language barriers are disadvantaged. We also asked how the CCTLDs or TLDs in general, which answered the survey, see the main barriers to adoption. And we identified three main fields. One are technical issues, the compatibility with browser and email services. Second, low public awareness. And also the habit to use ASCII, like English, is so rooted with the users that it’s very difficult to overcome this. One of the case studies we have on our IDN World Report website refers to the introduction of a top-level domain in Hebrew. And even though it was extremely also politically important for the country to introduce this, they reported back that it was extremely difficult to bring to the users that there is this possibility and that it can work in the domains in Hebrew. And the third barrier identified is also a modest level of promotion, both from registries and registrars, which is actually quite understandable. With the registries, it’s a bit better because they have, especially CCTLDs, in their DNA embedded support to local communities. But for registrars who are commercial companies, if there is not a business case, it’s difficult for them to find the resources for the support. This is to illustrate what you can find in the report. We have listed all the results, so you can see that indeed the majority voted for low awareness. Then also one of the sentiment questions was how organizations promote IDN. The most voted for marketing and also presence in events. These complete answers are there for all the questions we have asked. Shortly to the sources and methodology for CCTLD data, we are cooperating, apart from the forum, with the center and other regional organizations in Asia, Latin America, also Africa. And we also take the GTLD data from Mark Datysgeld and also from Zooknik. The mechanism or the season is that in January, the research team circulates two forms. One is about the IDN implementation in technical terms, and one is the sentiment survey. And then it’s complemented with the data, registration data, available via the regional registries organizations. And then also, as much as it is possible, the research team follows up with the CLTLDs to get a better understanding for the data. As our time is almost expired, and I still would like to give a space if there is some question, I will just refer you to also the recent articles and case studies in the website, including our colleagues from .cat, which have a very interesting way to promote and they put a lot of effort there. And also, I think the presentation will be available in the recording of the session, so some useful links. There is a lot of information on the ICANN website. There is also a recently formed CODI group. There were some changes in the universal acceptance steering group at ICANN, so you can have a look for further information. So I would like to finish with a call for action. If you have your IDN story from your country, please share it with us. We will be happy to publish it. Also, if you will be able to reach out so that we can address more CLTLDs, that would be very helpful. So thank you very much for your attention, and if you have a question to Esteve or to me, we still have space to answer it. Yes? I think, can I ask you? Thank you.

Tapani Tarvainen: Okay. Tapani Tajvainen from Electronic Frontier Finland, and also a non-commercial stakeholder group in ICANN. Have you been following the use of IDN top-level domains and ICANN’s policies in that regard? You mentioned Sámi, and in particular, there is a policy going on that looks like the Sámi letter Eng will be much less usable than some other special characters because nobody out there cares, basically. So that these very small minority languages and their characters would need someone to speak for them.

Regina Filipova Fuchsova: It’s a very good point, because there are also languages which cannot be put in written form, of course, and still deserve preservation. But since you mentioned Sámi, actually, I was in a session. It was in Tampere, so what, two years ago? Sandra can help me with Eurodic in Tampere, two years ago. And there was a session, and there was a young lady researcher who did research for the Sámi language. It was herself, I think not a Mavratan, but her family’s Mavratan. And there was obviously some movement of young people to preserve the language and raise awareness. So even though it will not become widely spread, I think especially via the young people, if they are interested in preserving this heritage online, we, as well, I can speak for the technical community, I think are obliged just to do our part of the work to enable this. But otherwise, you are right, some languages are even difficult to bring over to written form. So the content, actually, it starts and finishes with the content, of course, available. Do we have the mic already switched off?

Tapani Tarvainen: Okay, just to follow up. Sámi does not have a problem of not being able to write it. It just uses a bunch of special characters, and these are not well supported. That was the point.

Regina Filipova Fuchsova: But it’s a technical thing, right, to include them in the next version of the ID&A variant? Okay, we can talk about it. We had the last one in 2008, so high time to update. But thank you very much. Thank you very much for your attention. Thank you, Esteve, for your contribution. And please reach out to us. There is an email address. We will be happy to share and cooperate. Thank you.

Agreement points

Multilingualism as fundamental to digital inclusion and cultural preservation

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

Multilingual internet access is crucial for digital inclusion and preserving cultural heritage online, as illustrated by indigenous languages like Sámi

EU’s identity is fundamentally based on multilingualism with 24 official languages, and this priority extends into the digital realm

Summary

Both speakers strongly emphasize that multilingualism is essential for digital inclusion and maintaining cultural diversity online, with specific focus on preserving indigenous and minority languages in the digital space

Topics

Multilingualism | Cultural diversity | Development

Technical barriers significantly hinder IDN adoption and universal acceptance

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

Over half of registries don’t support Unicode addresses in email servers, and three-fourths don’t permit Unicode symbols as contact emails

Main barriers to IDN adoption include technical compatibility issues, low public awareness, and entrenched habits of using ASCII domains

Summary

Both speakers acknowledge that technical implementation gaps and compatibility issues are major obstacles preventing widespread adoption of internationalized domain names

Topics

Infrastructure | Multilingualism

Need for advocacy and support for minority languages in technical policies

Speakers

– Regina Filipova Fuchsova
– Tapani Tarvainen

Arguments

Multilingual internet access is crucial for digital inclusion and preserving cultural heritage online, as illustrated by indigenous languages like Sámi

Small minority languages face particular challenges with special character support in IDN policies and need advocacy

Summary

Both speakers recognize that minority and indigenous languages require dedicated advocacy and technical support to ensure their preservation and accessibility online

Topics

Multilingualism | Cultural diversity | Infrastructure

Similar viewpoints

Both speakers advocate for comprehensive universal acceptance where all languages and scripts should be equally supported across internet infrastructure and applications

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

Universal acceptance means all domain names and email addresses should work equally across all internet applications regardless of script or characters

EU policy and funding strongly support online multilingualism, requiring all official EU content to be available in every official language

Topics

Infrastructure | Multilingualism | Legal and regulatory

Both speakers recognize the need for broader distribution and support of multilingual internet infrastructure beyond current concentrated markets

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

The IDN market is highly concentrated, with top 10 TLDs accounting for 75% of all 4.4 million IDNs worldwide

EU policy and funding strongly support online multilingualism, requiring all official EU content to be available in every official language

Topics

Infrastructure | Multilingualism

Unexpected consensus

Connection between local language internet and resistance to internet shutdowns

Speakers

– Esteve Sanz
– Regina Filipova Fuchsova

Arguments

Local language internet could potentially reduce internet shutdowns because people would feel more ownership of the internet in their native language

Explanation

This represents an unexpected and innovative connection between multilingual internet access and internet freedom/human rights, suggesting that linguistic ownership could serve as a protective factor against government censorship

Topics

Multilingualism | Human rights principles | Freedom of expression

Overall assessment

Summary

Strong consensus exists among all speakers on the fundamental importance of multilingual internet access for digital inclusion, cultural preservation, and universal acceptance. All speakers agree on the technical barriers hindering IDN adoption and the need for advocacy for minority languages.

Consensus level

High level of consensus with complementary perspectives rather than disagreement. The speakers represent different stakeholder groups (registry operator, policy maker, civil society) but share aligned views on core issues. This strong consensus suggests potential for coordinated action on universal acceptance and IDN promotion, though implementation challenges remain significant due to technical and awareness barriers identified by all speakers.

Different viewpoints

Unexpected differences

Overall assessment

Summary

This discussion showed remarkably high consensus among speakers, with no direct disagreements identified. All speakers shared a common vision of multilingual internet access and universal acceptance.

Disagreement level

Very low disagreement level. The discussion was characterized by complementary perspectives rather than conflicting viewpoints. Regina provided technical and registry perspectives, Esteve contributed EU policy context, and Tapani raised specific advocacy concerns for minority languages. This high level of agreement suggests strong consensus within the technical and policy communities about the importance of multilingual internet infrastructure, though it may also indicate that more challenging implementation debates occur in other forums.

Partial agreements

Similar viewpoints

Both speakers advocate for comprehensive universal acceptance where all languages and scripts should be equally supported across internet infrastructure and applications

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

Universal acceptance means all domain names and email addresses should work equally across all internet applications regardless of script or characters

EU policy and funding strongly support online multilingualism, requiring all official EU content to be available in every official language

Topics

Infrastructure | Multilingualism | Legal and regulatory

Both speakers recognize the need for broader distribution and support of multilingual internet infrastructure beyond current concentrated markets

Speakers

– Regina Filipova Fuchsova
– Esteve Sanz

Arguments

The IDN market is highly concentrated, with top 10 TLDs accounting for 75% of all 4.4 million IDNs worldwide

EU policy and funding strongly support online multilingualism, requiring all official EU content to be available in every official language

Topics

Infrastructure | Multilingualism

Key takeaways

Universal Acceptance is critical for digital inclusion and multilingual internet access, requiring all domain names and email addresses to work equally across applications regardless of script or characters

The IDN market faces significant challenges with negative growth trends (-1% for ccTLDs, -5.5% for gTLDs) and high concentration (top 10 TLDs control 75% of market)

Technical barriers remain substantial – over half of registries don’t support Unicode addresses in email servers, and three-fourths don’t permit Unicode contact emails

Main adoption barriers are technical compatibility issues, low public awareness, and entrenched habits of using ASCII domains

EU’s multilingual digital policy strongly supports Universal Acceptance as aligned with its ‘united in diversity’ identity and 24 official languages

Multilingual internet access has broader social implications, potentially reducing internet shutdowns by increasing local ownership and preserving cultural heritage online

Resolutions and action items

Call for more TLDs to participate in the IDN World Report data collection to improve research comprehensiveness

Request for communities to share their IDN stories and case studies for publication on the IDN World Report website

Need to reach out to more ccTLDs to expand participation in the annual research surveys

Follow-up discussion needed on Sámi language character support in IDN policies between presenters and Tapani Tarvainen

Unresolved issues

How to overcome the low uptake of IDNs despite technical availability (88% of European ccTLDs support IDNs but adoption remains around 1%)

How to incentivize commercial registrars to promote IDNs when business case is unclear

Specific technical solutions for supporting minority language characters like Sámi letter Eng in IDN policies

How to bridge the gap between policy intentions and actual implementation of multilingual internet infrastructure

Strategies for increasing user awareness and changing entrenched ASCII domain usage habits

Suggested compromises

None identified

We’re very worried about Internet shutdowns, that, you know, despite all these commitments, they continue to be on the rise. And very intelligent people that know the local characteristics of these shutdowns, they tell us that if the Internet was in our local language, we would not see a shutdown. Because people would really feel the Internet as themselves, and politicians could not perform that shutdown.

Speaker

Esteve Sanz

Reason

This comment is profoundly insightful because it connects linguistic accessibility to political resistance and digital rights in an unexpected way. It suggests that when people have genuine ownership of the Internet through their native language, it becomes much harder for authoritarian governments to justify shutting it down. This reframes Universal Acceptance from a technical convenience issue to a fundamental tool for digital freedom and resistance to censorship.

Impact

This comment significantly elevated the discussion from technical implementation challenges to geopolitical implications. It prompted Regina to acknowledge this was a completely new perspective they hadn’t considered for future studies, showing how it opened new research directions. The comment transformed the conversation from focusing on European multilingualism to considering global digital rights and political freedom.

Have you been following the use of IDN top-level domains and ICANN’s policies in that regard? You mentioned Sámi, and in particular, there is a policy going on that looks like the Sámi letter Eng will be much less usable than some other special characters because nobody out there cares, basically. So that these very small minority languages and their characters would need someone to speak for them.

Speaker

Tapani Tarvainen

Reason

This comment is thought-provoking because it exposes a critical gap between the idealistic goals of Universal Acceptance and the harsh reality of implementation. It highlights how minority languages face systemic disadvantage not just from market forces, but from the very governance structures meant to support linguistic diversity. The phrase ‘nobody out there cares’ starkly illustrates how technical decisions can perpetuate linguistic marginalization.

Impact

This intervention shifted the discussion from celebrating progress in multilingual internet to confronting uncomfortable truths about whose languages actually get supported. It forced the speakers to acknowledge the limitations of current approaches and the need for active advocacy for minority languages. The comment introduced a note of urgency and activism that wasn’t present in the earlier technical presentation.

This might seem really low, 1%, but it’s actually in line with what our CCTLDs are experiencing. Also taking into account the technical and other issues, the uptake is not as it could be.

Speaker

Regina Filipova Fuchsova

Reason

While seemingly a simple statistic, this admission is insightful because it reveals the stark disconnect between the technical capability (88% of European ccTLDs support IDNs) and actual usage (only 1%). This honest acknowledgment of failure challenges the narrative of progress and forces a reckoning with why, despite years of development and policy support, multilingual domains remain largely unused.

Impact

This statistic served as a reality check that grounded the entire discussion. It shifted the focus from what’s technically possible to why adoption remains so low, leading to the detailed analysis of barriers (technical issues, low awareness, ASCII habits). It provided the empirical foundation that made the subsequent discussion of challenges more credible and urgent.

Overall assessment

These key comments fundamentally transformed what could have been a routine technical presentation into a nuanced exploration of the complex barriers to digital linguistic equality. Esteve’s insight about Internet shutdowns elevated the stakes from convenience to freedom, while Tapani’s intervention about Sámi characters exposed the gap between policy intentions and implementation reality. Regina’s honest acknowledgment of low adoption rates provided the empirical grounding that made these critiques meaningful. Together, these comments shifted the discussion from celebrating technical achievements to confronting systemic challenges, from European focus to global implications, and from abstract policy goals to concrete advocacy needs. The result was a much richer conversation that acknowledged both the importance of Universal Acceptance and the significant obstacles that remain in achieving true digital linguistic equality.

Could the relationship between Internet shutdowns and local language Internet usage be studied further?

Speaker

Esteve Sanz

Explanation

Esteve mentioned an interesting idea from Global South players that if the Internet was in local languages, politicians might be less likely to perform shutdowns because people would feel the Internet belongs to them. This concept wasn’t previously considered and could be valuable for future research.

How can more ccTLDs be encouraged to participate in the IDN World Report data collection?

Speaker

Regina Filipova Fuchsova

Explanation

Regina noted that the study has limitations due to the number of TLDs providing data and made a call for action to address more ccTLDs in their communities to participate in the study.

What are ICANN’s policies regarding IDN top-level domains, particularly for minority languages like Sámi?

Speaker

Tapani Tarvainen

Explanation

Tapani raised concerns about ICANN policies that may make certain characters (like the Sámi letter Eng) less usable than others, suggesting that small minority languages need advocates to speak for them in policy discussions.

When will the IDN variant tables be updated to better support special characters used in minority languages?

Speaker

Regina Filipova Fuchsova (in response to Tapani Tarvainen)

Explanation

Regina acknowledged that the last update was in 2008 and it’s ‘high time to update’ the IDN variant tables to include better support for special characters used in languages like Sámi.

How can IDN stories from different countries be collected and shared?

Speaker

Regina Filipova Fuchsova

Explanation

Regina made a call for action asking participants to share IDN stories from their countries, indicating a need for more case studies and examples to be documented and published.

Summary

This discussion at IGF 2025 focused on the challenge of regulating artificial intelligence in ways that are both ethical and innovation-enabling, examining whether current AI governance frameworks adequately balance these competing demands. The session brought together speakers from multiple regions to explore the tension between AI ethics as surface-level principles versus regulation grounded in ethical foresight that creates real accountability.

Alexandra Krastins Lopes emphasized that ethical foresight must be operationalized rather than theoretical, requiring proactive mechanisms built into organizational governance structures throughout AI systems’ lifecycles. She highlighted the need for global AI governance platforms that can harmonize minimum principles while respecting local specificities, noting that regulatory asymmetry creates concerning scenarios of unaddressed risks and international fragmentation. Moritz von Knebel identified key regulatory gaps including lack of technical expertise among regulators, reactive rather than anticipatory frameworks, and jurisdictional conflicts that create races to the bottom. He advocated for adaptive regulatory architectures and the establishment of shared language around fundamental concepts like AI risk and systemic risk.

Vance Lockton stressed the importance of regulatory cooperation focused on influencing AI design and development rather than just enforcement, noting the vast differences in regulatory capacity between countries. Phumzile van Damme called for binding international law on AI governance and highlighted the exclusion of Global South voices, advocating for democratization across AI’s entire lifecycle including development, profits, and governance. Yasmin Alduri proposed moving from risk-based to use-case frameworks and emphasized the critical need to include youth voices in multi-stakeholder approaches.

The speakers collectively agreed that overcoming simple dichotomies between innovation and regulation is essential, as safe and trustworthy frameworks actually enable rather than hinder technological advancement.

Keypoints

## Major Discussion Points:

– **Ethical Foresight vs. Surface-Level Ethics**: The distinction between treating AI ethics as a superficial add-on versus embedding ethical foresight as a foundational, operational process that anticipates risks before harm occurs and involves all stakeholders from design to deployment.

– **Regulatory Fragmentation and the Need for Shared Language**: The challenge of different countries and regions developing incompatible AI governance frameworks, creating regulatory gaps, jurisdictional conflicts, and the absence of consensus on basic definitions like “AI systems,” “systemic risk,” and regulatory approaches.

– **Inclusion and Global South Participation**: The exclusion of marginalized communities, particularly from the Global South and Africa, from AI governance discussions due to policy absence, resource constraints, and the dominance of Western perspectives in shaping AI regulatory frameworks.

– **Cross-Border Regulatory Cooperation**: The mismatch between globally operating AI platforms and locally enforced laws, requiring new approaches to international regulatory cooperation that focus on influencing design and development rather than just enforcement actions.

– **Innovation vs. Regulation False Dichotomy**: Challenging the narrative that regulation stifles innovation, with speakers arguing that safe, trustworthy frameworks actually enable innovation by providing predictability and building consumer trust, similar to other regulated industries like aviation and nuclear power.

## Overall Purpose:

The discussion aimed to explore how to develop AI governance frameworks that are both ethically grounded and innovation-friendly, moving beyond reactive approaches to create proactive, inclusive regulatory systems that can address the global, borderless nature of AI technology while respecting local contexts and values.

## Overall Tone:

The discussion maintained a professional, collaborative tone throughout, characterized by constructive problem-solving rather than adversarial debate. Speakers acknowledged significant challenges while remaining optimistic about finding solutions through multi-stakeholder cooperation. The tone was academic yet practical, with participants building on each other’s points and seeking common ground. There was a sense of urgency about addressing AI governance gaps, but this was balanced with realistic assessments of the complexity involved in international coordination and inclusive policymaking.

Speakers

**Speakers from the provided list:**

– **Vance Lockton** – Senior technology policy advisor for the office of the Privacy Commissioner in Canada

– **Alexandra Krastins Lopes** – Lawyer at VLK Advogados (Brazilian law firm), provides legal counsel on data protection, AI, cybersecurity, and advising multinational companies including big techs on juridical matters and government affairs. Previously served at the Brazilian Data Protection Authority and has a civil society organization called Laboratory of Public Policy and Internet

– **Deloitte consultant** – AI Governance Consultant at Deloitte, founder of Europe’s first youth-led non-profit focusing on responsible technology

– **Phumzile van Damme** – Integrity Tech Ethics and Digital Democracy Consultant, former lawmaker

– **von Knebel Moritz** – Chief of staff at the Institute of AI and Law, United Kingdom

– **Moderator** – Digital Policy and Governance Specialist

– **Online moderator 1** – Mohammed Umair Ali, graduate researcher in the field of AI and public policy, co-founder and coordinator for the Youth IGF Pakistan

– **Online moderator 2** – Haissa Shahid, information security engineer in a private firm in Pakistan, co-organizer for the Youth IGF, ISOC RIT career fellow

**Additional speakers:**

– **Yasmeen Alduri** – AI Governance Consultant, Deloitte (mentioned in the moderator’s introduction but appears to be the same person as “Deloitte consultant” in the speakers list)

# AI Governance Discussion at IGF 2024

## Balancing Ethics and Innovation in Artificial Intelligence Regulation

### Executive Summary

This session at the Internet Governance Forum brought together international experts to examine how to develop AI governance frameworks that balance ethical responsibility with technological advancement. The hybrid session featured speakers from Canada, Brazil, the United Kingdom, South Africa, and Pakistan, representing perspectives from government agencies, law firms, consultancies, and civil society organizations.

The discussion challenged the assumption that regulation necessarily stifles innovation, with speakers arguing that well-designed governance frameworks can actually enable technological progress. While participants agreed on the need for international cooperation and multi-stakeholder engagement, they offered different approaches to implementation, from flexible principle-based frameworks to binding international agreements.

### Session Structure and Participants

The session experienced initial technical difficulties with audio setup for online participants. The discussion was moderated by a Digital Policy and Governance Specialist, with online moderation support from Mohammed Umair Ali and Haissa Shahid from Youth IGF Pakistan.

**Key Speakers:**

– **Yasmeen Alduri**: AI Governance Consultant at Deloitte and founder of Europe’s first youth-led non-profit focusing on responsible technology

– **Alexandra Krastins Lopes**: Lawyer at VLK Advogados in Brazil, previously with the Brazilian Data Protection Authority

– **Phumzile van Damme**: Integrity Tech Ethics and Digital Democracy Consultant and former lawmaker from South Africa

– **Vance Lockton**: Senior Technology Policy Advisor for the Office of the Privacy Commissioner in Canada

– **Moritz von Knebel**: Chief of Staff at the Institute of AI and Law in the United Kingdom

### Key Themes and Speaker Perspectives

#### Moving Beyond Theoretical Ethics to Operational Implementation

Alexandra Krastins Lopes emphasized that ethical foresight “cannot be just a theoretical exercise” but must involve “proactive mechanisms” embedded in organizational structures. She argued for the need to operationalize ethics through concrete governance processes rather than treating it as an add-on to existing AI development.

Yasmeen Alduri reinforced this point by highlighting implementation challenges, noting that “if the developers cannot implement the regulations because they simply do not understand them, we need translators.” She advocated for moving from risk-based to use case-based regulatory frameworks that better account for how AI systems are actually deployed.

#### Regulatory Fragmentation and International Cooperation

Moritz von Knebel provided a stark assessment of current AI regulation, describing it as having “more gaps than substance” with “knowledge islands surrounded by oceans” of uncertainty. He identified critical shortcomings including lack of technical expertise among regulators and reactive rather than anticipatory frameworks.

Vance Lockton emphasized the importance of regulatory cooperation that focuses on “influencing design and development rather than just enforcement actions.” He highlighted the vast differences in regulatory capacity between countries as a critical factor in designing effective international cooperation mechanisms.

Alexandra Krastins Lopes proposed addressing fragmentation through “global AI governance platforms capable of harmonizing minimum principles on safety, ethics, and accountability while respecting local specificities.”

#### Global South and Youth Inclusion

Phumzile van Damme delivered a powerful critique of current AI governance processes, arguing that “there hasn’t been a proper democratization” of AI. She noted that while there may have been “somewhat of a democratization in terms of access,” true democratization requires addressing “the lifecycle” including “democratization of AI development, democratization of AI profits, and democratization of AI governance itself.”

Van Damme called for “binding international law on AI platform governance” and highlighted the systematic exclusion of African and Global South voices from AI governance discussions.

Yasmeen Alduri emphasized the exclusion of youth voices, noting that “multi-stakeholder approaches” often “forget about youth” despite young people having solutions and being disproportionately affected by AI systems.

#### Challenging the Innovation-Regulation Dichotomy

Multiple speakers challenged the narrative that regulation inherently stifles innovation. Moritz von Knebel argued that “you can have innovation through regulation,” citing examples from the nuclear and aviation industries where “they took off after we had safety standards in place.”

This perspective reframes regulation as potentially enabling innovation by providing predictability, building trust, and creating level playing fields that reward responsible development practices.

### Areas of Agreement and Disagreement

**Consensus Areas:**

– Need for multi-stakeholder engagement across civil society, private sector, and public sector

– Importance of international cooperation while respecting local contexts

– Inadequacy of current reactive regulatory approaches

– Need for adaptive rather than static frameworks

**Different Approaches:**

– **Regulatory mechanisms**: Flexible principle-based approaches versus binding international law

– **Framework design**: Risk-based versus use case-based categorization

– **Implementation focus**: Early-stage development influence versus enforcement cooperation

### Questions and Audience Interaction

The session included questions from online participants about practical implementation challenges and the role of friction between tech companies and regulators. Yasmeen Alduri noted that such friction is expected and that “democratic states specifically live from this friction,” while emphasizing the need for constructive dialogue.

Participants also discussed the challenge of balancing innovation incentives with necessary safety regulations without creating regulatory arbitrage where companies seek the most permissive jurisdictions.

### Practical Recommendations

Speakers proposed several concrete steps:

– Establishing global AI governance platforms for harmonizing minimum principles

– Creating independent technical advisory groups and capacity building programs for regulators

– Developing international dialogues to establish consensus on key AI definitions and terminology

– Creating spaces for meaningful co-creation between different stakeholders

– Shifting narratives around AI from inevitability to focus on desired societal outcomes

### Unresolved Challenges

The discussion highlighted several ongoing challenges:

– How to achieve definitional consensus across different regulatory traditions and cultures

– Addressing vast disparities in regulatory capacity and resources between countries

– Developing practical mechanisms for meaningful Global South and youth inclusion

– Implementing effective cross-border enforcement while respecting national sovereignty

### Conclusion

The discussion demonstrated both the potential for collaborative AI governance and the complexity of implementation. While speakers found common ground on fundamental principles, they revealed significant challenges in translating shared values into coordinated action. The session’s emphasis on inclusion and its challenge to the innovation-regulation dichotomy provide important foundations for future AI governance efforts, though substantial work remains to develop practical, implementable frameworks that serve global needs.

Moderator: I hope everyone can hear me. Hello, everyone. Good morning and welcome to this session, Digital Dilemma, AI Ethical Foresight versus Regulatory Ruling. I’m a Digital Policy and Governance Specialist, and this is a conversation that I’ve been looking forward for many weeks. Today is the final day of IGF 2025, and it’s fitting that we’re closing with one of the most important global governance challenges. How to regulate artificial intelligence in a way that’s both ethical and enabling? Is that even possible? As the awareness of AI’s power has risen, the dominant response has been to turn to AI ethics. Ethics, simply put, is a set of moral principles. We’ve seen governments, companies, and institutions roll out principles, guidelines, and frameworks meant to guide responsible AI development. But here is the problem. When ethics is treated as a surface-level add-on, it often lacks the teeth to create real accountability or change, and that’s where regulation grounded in ethical foresight becomes essential. We’ll be unpacking the challenge today through insights from multiple regions and perspectives. I’m joined in person by three incredible speakers, Yasmeen Alduri, AI Governance Consultant, Deloitte, Alexandra Krastins Lopes lawyer at… and VLK Advogados, and Phumzile van Damme, Integrity Tech Ethics and Digital Democracy Consultant, who will be joining us in just a short while. We also have a vibrant virtual component, my colleagues, Mohammed Umair Ali and Haissa Shahid, are the online moderators. Umair, over to you to introduce yourselves and our online speakers.

Online moderator 1: Hello, everyone. Hello, everyone. Just mic testing. Am I audible? Yes. Great. So welcome, everyone. Welcome to our session. I’m Mohammed Umair Ali. I’ll be serving as the co-host and online moderator for this very session. Briefly about my introduction. I’m a graduate researcher in the field of AI and public policy. I’m also the co-founder and the coordinator for the Youth IGF Pakistan. And by my side is Haissa Shahid. Haissa, can you turn on your camera?

Online moderator 2: Hello, everyone. Sorry, I can’t turn on the camera right now. But hello from my side. I’m Haissa Shahid from Pakistan. And I’m also the co-organizer for the Youth IGF. Along with that, I’m serving as an information security consultant. I’m also the co-founder and the coordinator for the Youth IGF Pakistan. And by my side is Haissa Shahid. Haissa, can you turn on your camera? Hello, everyone. Sorry, I can’t turn on the camera right now. But hello from my side. I am serving as an information security engineer in the private firm in Pakistan. And I’m also an ISOC RIT career fellow this year. So, I’ll bring forward the session ahead. Thank you.

Online moderator 1: And moving towards the introduction of our speakers, so we are joined by speakers from the United Kingdom and Canada. To start with Mr. Vens, who is the senior technology policy advisor for the office of the Privacy Commissioner in Canada. And Mr. Meritz from the United Kingdom, who is the chief of staff in the intelligence division of the ICT division. Please welcome Mr. Meritz. Thank you so much for having me today. Hello, everyone. Mr. Meritz is the chief of staff at the Institute of AI and Law. Welcome aboard, everyone. Over to you, Taiba. Taiba, am I audible?

Moderator: Thank you so much for letting me know. I’m going to repeat myself. We’ll begin with a round of questions for each of our five speakers. One question each, about five minutes per speaker towards the end. We will be taking pictures, speakers with the audience, so please don’t run off too quickly at the end. Alexandra, I’ll start with you. So many of the AI frameworks we see today cite ethics as a very important aspect. But when you look at it a bit closely, ethics often feels like an add-on rather than like a foundational principle. From your perspective, is that enough? And more importantly, what does it look like to truly embed ethical foresight into governance structures so that it shapes the direction of AI development from the very start?

Alexandra Krastins Lopes: Great, thanks. It’s an honor to contribute to this important discussion. And while I have a proudly funded civil society organization called Laboratory of Public Policy and Internet, and served for a few years in the Brazilian Data Protection Authority, today I speak from the private sector perspective. I represent Felicade Avogados, a Brazilian law firm where I provide legal counsel on data protection, AI, cybersecurity, and advising multinational companies including big techs on juridical matters and government affairs. And when we talk about embedding ethical foresight into AI governance, we’re not talking about simply including a list of ethical principles in a regulation or policy document. We’re talking about building a proactive and ongoing process, one that is operational, not theoretical. Ethical foresight means anticipating risks before harm occurs. So it requires mechanisms that allow organizations to… to ask the right questions before AI systems are deployed. Are these systems fair? Are they explainable? Do they reinforce historical discrimination? Are they safe across different contexts? And this ethical foresight must be built into organization governance structures, including AI ethics committees, impact assessments, internal accountability protocols that operate across the entire lifecycle of AI systems. It also requires clarity about roles and responsibilities within organizations. Therefore, ethical foresight is not just the job of the legal or the compliance team, it needs engagement from product designers, data scientists, business leaders, and most importantly, it needs board-level commitment. At the same time, foresight must be context-aware. Many global frameworks are still shaped primarily by Nord countries’ perspectives, with assumptions about infrastructure, economic and regulatory capacity, and also risk tolerance that do not necessarily reflect the realities of the global majority. So the regulatory debate should be connected to social context. In Brazil, for example, historical inequalities, specific challenges for entrepreneurship, and limited institutional enforcement power. Besides that, embedding ethical foresight in governance requires abstract principles. That is, by setting minimum requirements rather than imposing a one-size-fits-all model. And tools can be adapted to specific conditions, such as voluntary codes of conduct, soft law mechanisms, AI ethics boards, and flexible risk-based approaches. Additionally, national policies may have different strategic goals, such as technology development, This regulatory asymmetry creates a highly concerning scenario of unaddressed ethical and social risks, international regulatory fragmentation, and direct normative conflicts. Embedding AI foresight requires international coordination with minimum safeguards, promotion of technological progress, investment, and global competitiveness. Considering this scenario, the proposal I bring for reflection today is the need to establish global AI governance platforms capable of harmonizing minimum principles on safety, ethics, and accountability while respecting local specificities. Besides that, incentivize tools such as voluntary codes of practice, which allow business to anticipate and mitigate risks without restricting innovation. Additionally, there should be parameters for cross-border regulatory cooperation for enforcement and accountability, since jurisdictional fragmentation makes enforcement difficult, especially when platforms operate globally, but accountability is local. In my experience advising companies, I’ve seen that while they are willing to act responsibly and ethically, international cooperation and legal certainty are essential for that to happen. Thank you.

Moderator: Thank you so much, Alexandra, and as she mentioned, ethical foresight is not just the job of a legal team, but truly a multi-stakeholder process. Now we’ll turn to our online speaker, Moritz, if you’re there, you’ve been keeping a very close view on how AI regulatory frameworks are evolving and also where they’re falling short. From where you sit, from your perspective, what are some of the key loopholes or blind spots in current AI regulation and what would it take in terms of practical terms to close these gaps and build a governance model that builds digital trust and resilience? Over to you.

von Knebel Moritz: Yeah, thank you and thanks for having me. People have often asked this question, what are the regulatory gaps that we see? And I think that assumption already is a bit flawed and mistaken because that would assume that around these gaps we have well thought out systems and processes and frameworks. And I don’t think we have. And so I will employ a metaphor used by a former professor of mine who said that he does not have knowledge gaps because it would be just too large. But he has knowledge islands and those are surrounded by a lot of ocean that we don’t know about. And then if those islands are well connected enough, then you can travel back and forth and it still kind of works. But there are more gaps than there is actual substance. So with that caveat, let’s look at where those gaps are clearest or where the ocean is deepest, so to speak. And I think on the domestic level, I would identify two big items. One is a lack of technical expertise. And so regulators often lack the deep technical understanding that is needed to effectively oversee those rapidly evolving AI systems. That’s especially true in countries that have historically not had the kind of capacity and resources to build up that infrastructure. And those institutions that then in turn creates a reliance on industry for setting standards for self-reporting, for self-assessing the safety of their models. And that makes it very difficult to craft meaningful technical standards on the. on the political level. That’s one of these very, very deep, deep sea areas. Another one is that we often see reactive frameworks. So current approaches largely respond to known harms and try to track when a harm occurs. But they do little in terms of anticipating emerging risks. And if you think that that is where a huge amount of the risk in the future is going to come from, then frameworks need to be adaptive to that. And the pace of AI development, which proceeds at a breakneck speed, consistently outstrips the pace at which regulatory systems can adapt, which is a huge challenge. On the international level, I think there’s one big item that then feeds into other problems, which is jurisdictional conflicts or races to the bottom. And then that turns into sometimes unjustified, sometimes justified fears of overregulation. And so different national approaches, but also international approaches like the one that the EU has taken, create a lot of complexity for compliance and regulation. And China’s governance system, which focuses on algorithms, differs fundamentally from the EU, which is more risk based. The US, the UK have emphasized innovation friendly approaches, and that creates a regulatory patchwork that is difficult to navigate, but also creates room for regulatory arbitrage. So similar to how you would move your country to your company to a country that has very low tax rates, you might do the same when it comes to regulation. And so that then creates incentives also for countries to to water down and then weaken their regulation. All of this is amplified by the fact that we just do not know enough. So a lot of the terms that are used in the concepts like risk, systemic risk, are insufficiently defined and there’s no consensus on what that actually means. So on that maybe a bit gloomy note, I’ll move on to what we can take to close these gaps or rather to to build bridges between those. that we already have with the UAI Act and some other frameworks that exist around the globe. One thing is I think we need adaptive regulatory architectures, which means that rather than having static rules, we need frameworks that can evolve with technology and rather quickly. And so that could mean that you have principle-based regulations, and then technical implementation standards can be updated through streamlined processes in secondary regulation. You could also have regulatory sandboxes where you try out new approaches and see what works, and have very quick feedback loops, which maybe governments aren’t historically good at, but are definitely getting better at as we speak. There’s a general need for capacity building. So I’ve talked about how expertise is lacking in governments often, but it’s also lacking elsewhere. So creating independent technical advisory groups could be useful that include the perspectives of different people. And that also means that we need inclusive and novel approaches to engage stakeholders. We need dialogues and diplomatic efforts. So I’ve said that some terms are just not sufficiently defined. And so without a shared language, the landscape will remain fragmented and incentives to cut corners will remain. So we see some work on this with international reports coming out and the international dialogues on air safety, for instance, but much more of this is needed to establish consensus on the key questions, because only then can we start to think about regulatory frameworks. And one last thing, because I do work on AI, so I have to flag this. There might also be ways to leverage AI for these processes. And I think relatively little work has gone into this and much more is needed. There are many more gaps and many more solutions that we could talk about, but I’ll leave it at that for now. And I’m happy to receive questions about any or all of these in the Q&A.

Moderator: Thank you, Moritz. I have a question. You mentioned a shared language. Can you elaborate more on that? Because it sounded really interesting.

von Knebel Moritz: Yeah, so this runs the gap. It basically spans from the very fundamental, what is the risk? What is an AI system? Luckily, the OECD provides a definition that most people are happy with. But when it comes to, I’ll give you a concrete example, the definition of systemic risks in the European Union under the AI Act, this will influence a lot on how future AI models are governed and deployed and developed. And so since that is the case, it really matters what people think of as systemic risk. And it’s not a field that has centuries of research behind it, but rather decades. And so some people understand very different things under systemic risk as others do. And so unless you have some consensus on this, it’ll be very, very difficult. That’s on the sort of fine grained regulation and a specific legislative framework. But there’s also high level conversations about what are even the most important risks and to try to create consensus on this. Again, different cultures, different countries see AI in a very different way. And if you consider that, that makes it difficult to cooperate and collaborate internationally. And so to establish a shared language around what does AI risk mean? And I’m specifically focused on risk here, but the same goes for benefits. What are the benefits that we care most about? That will be needed. And that, again, also touches on the technical expertise, because a lot of this requires a kind of technical knowledge and you need to know what accountability is and what adversarial robustness is of an AI model. So that further complicates things.

Moderator: Thank you so much, Moritz. Very enlightening. And the main key takeaway I could see was that different countries see AI in a very different way. And I think we really need to explore this further. Vance, I’m going to go to you now. Given how borderless technology is, there’s a real challenge around jurisdiction and enforcement. And I know Moritz also touched upon it. In a way, that platforms operate globally, but the laws don’t. What opportunities do you see for cross-border regulatory cooperation, and how can regulatory cooperation help tackle jurisdictional conflicts and enforcement barriers, especially for digital platforms that may be lightly regulated in one country, but highly impactful globally? Over to you, Vance.

Vance Lockton: Sure, thanks. So, it’s a very interesting question when we get into this idea of regulatory cooperation, because, I mean, as Moritz has been flagging, not only are the understandings of artificial intelligence difference across countries, there’s such a difference on what regulatory frameworks actually exist, and what can be applied, that once we’re into the realm of enforcement, certainly, there are some countries that have attempted to do shared, make shared enforcement efforts, but I don’t think those are necessarily where we need to focus our efforts. They’re not necessarily going to be the most effective way to address some of these issues. I think what really needs to be happening, when we’re talking about regulatory enforcement, rather than shared enforcement actions, it really has to be, how can regulators have more influence over the design and development and deployment decisions that are being made when these systems are being created or adopted in the first place? And that’s something that a lot of organizations are starting to try to, try to wrap their heads around. I look at a document like the OECD has a framework for anticipatory governance of emerging technologies. I don’t have any particular connection to it. The OPC, my organization, doesn’t have it particularly endorsed this document, but it just kind of creates this useful framework for me of saying, you know, the elements that they set out in this framework of what this regulatory governance needs to look like, saying, you know, it needs to establish guiding values, have strategic intelligence, bring in stakeholder engagement, agile regulation, international cooperation. Again, I don’t want to walk through this framework in particular, but I do think there are useful pieces that come out of that. Because from a regulatory perspective, one of the challenges that we always face is that we don’t really get to set the environment in which we work. Generally speaking, regulators are not going to be the ones who are drafting the laws that they’re overseeing. We may be able to provide input into their drafting, but by and large, we’re going to be handed a piece of legislation over which we have to have oversight. So, there are elements, but even within that, though, there are elements of discretion as far as what kinds of resources can be applied or what strategies can be applied to particular problems. Or there are often, in a lot of statutes, there will be considerations with respect to appropriateness or reasonableness. I think that’s going to be a critical piece for AI governance, for particularly the challenge that you’ve set out, this idea of international impacts of AI systems, where, you know, frankly, as a regulator in a Western nation, we aren’t necessarily going to be able to have that direct visibility, that direct insight into the impacts of that system on the global majority. So we aren’t necessarily going to have, again, that full visibility into what safeguards you need to be pushing for, or what appropriate purposes or what reasonableness actually means in the context of these AI systems, when we aren’t seeing what the true impacts of these systems are. So having those kind of dialogues amongst regulators to share that cultural knowledge is going to be a critical piece of this going forward. And, you know, just the sheer regulatory cooperation of understanding amongst regulators who has what capacities in AI, as Moritz flags, for a lot of regulators, there isn’t necessarily going to be that technical expertise. And that can be quite understandable. You know, for a lot of newer data protection authorities or some of the less resourced authorities, you might have a dozen staff for the entire regulatory agency. And that’s not a dozen staff on AI. That’s a dozen staff covering privacy or data protection as a whole. And obviously, data protection isn’t the only piece of AI regulations. It’s just kind of the piece I’m most familiar with. That’s why I’m kind of focused on it. So, you know, you might have a dozen staff. Canada has about 200 staff, but we have a dedicated tech lab that’s designed to be able to kind of get into these systems. It can take them apart and really understand what’s happening kind of under the hood of a lot of these AI systems. Then you look at something like the United Kingdom’s Information Commissioner’s Office, which has well over a thousand staff. So, well over a thousand staff and a lot of programs set up for ways that you can kind of have more innovative engagements with regulators around things like regulatory sandboxes and things along those lines. So, again, being able to understand amongst regulators who has what capacity within AI is going to be a critical piece. And I think having that shared understanding of who can do what and what the true risks and benefits are going to be is such a critical piece because, you know, I think one of the biggest challenges for me that I’m kind of seeing from a regulatory perspective is – again, from a Western regulatory perspective, I’ll say is that there’s this narrative that’s coming out that AI is such an essential piece of future economic prosperity of a nation or even security, future security of a nation. That there’s this idea that any regulation that creates restrictions on the development of AI is a threat as opposed to this opportunity or as opposed to this necessary thing to get to responsible innovation. And so being able to have that real understanding of what the potential risks are, what the potential harms are, and realistically what benefits we’re trying to achieve, that are not simply just, again, making stock prices for a handful of companies go up or making overall GDP of a country raise.

Moderator: Vance, I’m so sorry. You have 30 seconds left.

Vance Lockton: Yeah, no problem. Not a problem. So, again, we need to have that ability to counter that narrative. And, you know, within countries, regulators are getting better at having that cooperation amongst various sectors. So we are working towards having better cooperation internationally and finding those soft mechanisms, finding those ways to have influence over design and development. Again, it’s a work in progress, but my overall message is going to be we just need to reframe regulatory cooperation away from enforcement cooperation and to finding opportunities to influence that design and development. And I’ll stop there.

Moderator: Thank you so much, Vance. Phumzile, thank you so much for joining us. I’d love to talk to you about power and participation, shifting the lens to inclusion, which is, you know, a very important part of artificial intelligence. So we see that AI governance conversations are often dominated. by a handful of powerful perspectives. In your view, whose values are shaping the way AI is being regulated now, and what would it take to ensure that voices of marginalized or under-representative communities are not just included but also influence decision-making?

Phumzile van Damme: Thank you. I would just like to start with what I view as the ideal on AI governance. I think previous speakers have kind of highlighted a major challenge in that there is one, a lack of shared language. There is the idea that AI, like all technology, knows no jurisdictional borders. It’s global. So what I think the ideal would be is some form of international law around AI governance. And through that process, perhaps run through the UN, there is the ability to get a more inclusive process that includes the language, the ideals, the ethical systems of various countries. So I think that is the direction we need to be going. I think the experience from kind of over the last years with social media and the governance around that and the difficulty with different countries, either in some instances not having any laws, regulations or policies in place and others doing that, is that there is a lack of proper oversight. So I think it’s an opportunity now for international law to be drafted around that. But as it stands right now, outside of what I think is the utopia, my utopia, what I think would be the great thing to have around AI governance, I think there are two challenges of inclusion right now. One is that in the global south, particularly in Africa, there is policy absence. I know in many African countries, AI is just not a consideration, and that’s not because, you know, it’s just something that’s not viewed as important. There is a conflict with bread and butter issues. Most of these are countries that are more concerned or more, you know, they’re more focused on those types of issues. So I think because of that, then there is a exclusion in those discussions because, you know, that’s just a process that hasn’t taken place in those countries. So I think I’d like to see more African countries, countries in the global South, including themselves in those conversations by beginning to begin the process of drafting those policies, crafting those laws. And I always say that it may appear incredibly complex, but I think there is a way of looking at what other jurisdictions have done and kind of amending those laws to reflect local situations. I don’t think it’s a situation where there needs to be a reinvention of the wheel. I used to be a lawmaker, and I don’t want to say this in a polite way, but there’s a bit of, I don’t want to say that tech is, I don’t know, tech literacy is not the right phrase, but it’s just not something that’s…

Moderator: You can be candid. It’s okay.

Phumzile van Damme: Yeah. It’s just, it seems like a very difficult topic to handle, and it’s just avoided. So just to kind of encourage that it’s not that difficult, there are frameworks in existence that can be amended to fit the requirements in each country. So one, so there’s that exclusion. And I think there’s… The theoretical misnomer around what has been seen as a democratization of AI, and I think that is the wrong phrase to use, there hasn’t been a democratization. While there may have been somewhat of a democratization in terms of access, in terms of access to tools, gen AI tools, particularly like ChatGPT and all of those, there hasn’t been a proper democratization. So if we’re going to talk democratization and AI governance, we need to talk about it through the life cycle. So there needs to be democratization, I wrote this down, not only democratization of AI use, democratization of AI development, so that in the creation of AI tools, there’s a really diverse voice is included in those design choices, and large language models that there is someone at the table that says, you need to look at African languages, you need to look at languages from different countries, there needs to be a democratization of AI profits. You know, it can’t be a situation where countries are merely seen as sources of income, and there’s no way to kind of share profits, so there needs to be a democratization of AI profits, and the democratization of AI governance itself. So the way I kind of summarize it, I think there needs to be a deliberate effort by the global South, by African countries, by other countries, to insert themselves forcefully into the discussion, and that requires them beginning those processes where it hasn’t begun, and indeed for there to be a deliberate inviting of those voices to say, come take part in the discussions, and the discussion is not only at the stage where the technology is being deployed from, but through the entire, to the entire lifetime.

Moderator: Thank you. Very insightful. I actually have a question, but we’re running a bit short on time, so I’m probably going to leave it towards the end. But I really like the way that you said that the inclusion of gender inclusion or diversity of perspectives is not very difficult to do so, but it just seems very difficult to do so in real world in making sure that regulations or frameworks actually include a part of that as well. Thank you so much. Yasmin, we’re going to move on to you. We’re going to wrap up this round. We’ve talked a lot about the challenges, what’s not working, but let’s imagine what could. If you had a blank slate to design an ideal AI regulatory framework, one that’s forward-looking, ethically grounded and innovation-friendly, I mean huge to ask what are the three most essential features you would include?

Deloitte consultant: Good morning everyone. My name is Yasmin Alduri. I’m an AI governance consultant at Deloitte and I’m the founder of Europe’s first youth-led non-profit focusing on responsible technology. So we’re basically trying to bring young voices to the responsible tech field and have them basically share their ideals, their fears and their ideas on how to make this world a better place, let’s just say like that. I’m really happy van Damme actually brought up the point of international law because yes, in an ideal world we would all get along and in an ideal world we would have some kind of definition on AI governance, specifically on the international law, but there are two aspects that are really, really, I wouldn’t say problematic but that make this utopia hard to achieve. I wouldn’t say It’s not a possibility to achieve it, but it’s hard to achieve it. And that’s the first part is definitions. We need to get to a point where everyone defines the same aspects the same way. So just the definition of AI systems is a huge, huge issue. We saw this with the European laws already, with the EU AI Act. Then the second part, which is also an issue, is are countries actually upholding international law already? And if we look at the past years, we will see that in a lot of countries we see a huge increase in countries actually not upholding international law. So the first part would be to actually question and to basically push for more accountability from countries. So this is one aspect. I also wanted to bring up one point because I love the fact that also Saba Tiku Beyene brought up the aspect of inclusion. We need to set a base, which is infrastructure. If we don’t have infrastructure in different countries who are lacking the access to AI, we’re not able to democratize the access to AI. So this would also be a base for me. Okay, let’s come to the actual question. Ideal frameworks. As I said, in an ideal world, we wouldn’t have an approach that is so reactive, as Moritz called it earlier, but we would have one that is adaptive. So in this case, what we have right now is a risk-based approach. So we basically look at AI systems and we categorize them in different risks or straight up just prohibit them. And while I do see why we’re doing that, the ideal form would actually be a use case framework. So we actually look at AI systems and we look at them, how they’re used and in what sectors they’re used. The idea behind it is that AI systems, the same AI system, can be used in different use cases and in different ways, which means it can have different harms to different stakeholders. And the idea here is to make sure that we can actually use these systems. AI systems but we make sure that we really do uphold all the different scenarios that could happen and a use case approach would actually make this easier for us. It would also make it easier to keep and check the impact assessments that you have to do as someone who’s actually implementing those laws. That brings me actually to the second part which is we need frameworks that are actually implementable and understandable. So what exactly do I mean? At the end of the day those regulations and laws are being implemented by the people who are developing AI systems because those are the ones who are actually building it and those are the ones who have to uphold it. So if the developers cannot implement the regulations because they simply do not understand them, we need translators. So people like me or Alex who like go down and like try to explain or be the bridge between tech and law. But if we in an ideal world we would have developers who already understand this. We would have developers who would be able to implement each aspect, each principle in a way that is clearly not only defined but understandable. So this is the second part. The third part and I know that some of you will laugh already because you have heard this so many times over the past days. We need to include all stakeholders, multi-stakeholder approaches. So civil society, private sector and the public sector need to come together. Now here’s the thing where I disagree with most multi-stakeholder approaches. We talk a lot about including all stakeholders yet we forget to include the youth. And this is one of the biggest issues I see because in my work within the responsible technology hub I see not only the potential of youth but I see that they mostly even understand the technology better and they know how to implement it better. So we need to include them to not only future-proof our legislation but to make sure that they’re included in means of okay these are their fears, these are their not only ideas but these could be possible solutions. And from my work I can tell you that youth, young people, and young professionals have a lot of ideas for solutions that we’re facing, they’re just not heard. So we need to give them platforms. We need an older generation that leads by example but also gives platform. We need a young generation that refreshes this discourse every time. And we need spaces that are spaces of real co-creation. So these kind of discussions that we’re having right now, they’re a great base. But we need spaces where we actually create, where we have enough space to talk, to discuss, but at the same time to bring different disciplines and different sectors together. So in my work specifically, I see this working actually very well. So having a space where academia comes, where the public sector comes, where the private sector comes, and where we have intergenerational solutions for issues. And we’re actually building solutions together. So with that in mind, those would be the three aspects of, I would say, the ideal AI governance framework for me specifically.

Moderator: Right on time. Thank you so much to all our speakers. So many valuable takeaways. Now we open it up to the floor. The mics are on either side of the room. Please introduce yourself. And if you’re joining virtually, feel free to post in the chat or raise your hand. Omer, do we have any questions?

Online moderator 1: Yes, absolutely. We do have questions. I have received questions in the direct message and the chat box, but do we have any questions from the on-site participants before we move on to the online participants?

Moderator: We’re warming up here. So you know, start with the question. Right, so one of the questions that I received is,

Online moderator 1: and I think it might be for Moritz to, you know, maybe answer this better. The question is that, just like the concept of internet fragmentation, do you see that we are heading towards an era of AI fragmentation due to absence of consensus-based definition, absence of shared language, and if that is the case, how do you suggest to move forward with it?

von Knebel Moritz: Yes, I do think that we do see fragmentation, but I also think that it’s maybe a little bit reductive. I think this often gets portrayed as a fragmentation alongside country lines or jurisdictions like EU, US, China, whereas there are often splits between different factions and their interests and what they see as the ideal governance system, and all these factions have input. And so I see a fragmentation happening on multiple levels, so it goes beyond different countries and or jurisdictions. I am fairly concerned about fragmentation because, as I said, it creates the wrong incentives to race to the bottom. It doesn’t build a great base for trust between different partners, again, different countries, but also different sectors, the non-profit sector, industry, academia, government. So yes, I see this as a concern. In terms of what can be done, I think it goes back to identifying the areas where people already agree on and then building up from there. There’s never going to be perfect agreement and total unison on these things, but I think sometimes there’s more room and more overlap people are willing to give credit for. So in the foreign policy or diplomatic domain, Track 1.5, Track 2 dialogues have been pretty successful, I think, in identifying these areas of consensus. And yeah, I mean, at the end of the day, it’s also events and fora like IGF, where people get together and hopefully get to, at some point, establish a shared language or at least hear what other people, I think sometimes people are also too ambitious here, right? They want everybody to speak the same language and cultural anthropologists and other people will tell us that that is not actually what we should be aiming for, right? It is okay for people to speak different languages. But I think if you dial back your ambition and say, no, we don’t have to have a complete agreement, the first step is hearing how the other side sees things. Then I think having more of these conferences and as other people have said, diverse representation of voices at these conferences and events can be very useful in charting a path forward towards a shared language. It’s an iterative, it’s a painful process. It’s not going to come overnight, but it doesn’t mean that we shouldn’t invest energy in it.

Online moderator 1: All right. So, there is another question here and it says that, right. So, yeah, can you, can any of the speakers point towards any exemplary AI policy that balances innovation and ethics? I understand this is quite a broad-ended question, but if anyone would like to take that up.

Moderator: Anyone from the onsite speakers would like to take this question?

Deloitte consultant: Perfect. Let me take this off first. So, can I point any regulation out there that brings in good governance and with the news lately, so even there we have still discussions on how to implement specific laws and there is still a lot of room. So for example with the Middle East, we barely have any official AI governance there. We see a trend of discussions. We saw this at the last IGF where a lot of Middle Eastern and North African states came together and started having a discussion but this discussion relied more on principles rather than actual frameworks that could be implemented. So there is still, for whoever who asked this question, there is still a lot of room to actually co-create and to bring regulation into the market.

Moderator: Great, Harsha, I believe you got a question from the chat as well, do you want to go?

Online moderator 2: Yeah, we have another question. So Ahmed Khan asked, there has been a recent trend of boringness and putback from tech companies against what they call excessive regulation which restricts innovation and goes against development. So is there a way forward with Harmony or are we going to see a similar friction in the AI space as we have seen in social media and tech space versus government regulation?

Moderator: Anyone from the on-site speakers want to take this ahead or have a comment to make on this? Going to go online as well. Moritz or Vance, do you want to answer this question? Harsha, can you please repeat the question again, though? We didn’t understand it quite as the connection got a bit fuzzy.

Online moderator 2: So the question is, is there a way forward with Harmony, or are we going to see a similar friction in the AI space as we have seen in social media and tech spaces versus government regulations?

Moderator: So what I understand is that, you know, in the past we’ve seen that there has been some friction between platform regulation or platform governance as well. Are we going to be seeing a similar sort of friction in AI framework, AI regulation as well, as we move forward in the future? Do you think so? You don’t think so? You know, just a one word answer would be great as well.

Alexandra Krastins Lopes: Let me answer that one. I hope not. That’s why we’re here, having this multi-stakeholder discussion, so we can achieve some kind of consensus. And also complementing the last question about different kinds of legislations and which one would provide safeguards and would not hinder innovation. I believe that the UK has been taking a good approach by not regulating with a specific bill of law. There is no bill of law going on right now. They are letting the sectoral authorities handle the issues and also they did not define the concepts, so it won’t be a closed concept. So it can let the technology evolve and we can evolve concepts as well. So I hope that, I believe that a principle-based approach and not a strict regulation would be the best approach and that would prevent the friction we’re talking about.

Moderator: Okay, great, perfectly concise answer and good thing too because we’re running a bit short on time. So we’re going to move towards the final stretch. I’d like to hear one last thought from each speaker. Please restrict your answer to one sentence or two sentences at max. Yasmin, we’re going to start with you. What is the most urgent action we must take today to align AI governance with ethical foresight?

Deloitte consultant: To give a quick, quick answer to the last question about the friction between tech companies and regulators, yes, we see the friction. Yes, there will be more friction and I would be surprised if we didn’t have it, if specifically in democratic states specifically live from this friction. So we will see it and hopefully we will find a common ground to discuss these regulations. But what I do think is critical and what we need specifically as a society regardless of government regulation or tech regulation or whatnot, we need to make sure that we’re critically assessing what we’re consuming and we need to make sure that we critically assess what we say, what we perceive and what we create because what we’re seeing right now is very worrisome in means of how we’re using specific AI systems without questioning their outputs, we’re publicizing them without questioning what whether the contexts are right or not. So I believe one of the most important things right now as citizens is to make sure that we’re bringing critical thinking back again.

Moderator: Phumzile, would you want to go next?

Phumzile van Damme: Yeah, I’m gonna restrict it to one sentence. I’ve already gone into it. Binding international law on platform governance, AI platform governance.

Moderator: Perfect. Alexandra, would you want to go next?

Alexandra Krastins Lopes: I would like to leave you with a key action point that we must move from ethical intention to institutional implementation. I think ethical foresight cannot remain a theoretical aspiration or a paragraph and a regulation or a corporate statement. We need to build core structures of governance within companies and organizations in general.

Moderator: Rightly so. Ethical foresight should not be just a corporate statement. Thank you so much, Alexandra. And we’re going to go online to our online speakers. Vance, if you want to go ahead with two or three sentences to this question, what is the most urgent action we must take today to align AI governance with ethical foresight?

Vance Lockton: I’d say it’s to shift the narrative around AI away from AI being either an inevitability or a wholly necessary piece of future economies, to think about… what outcomes we want from future societies and how AI can build into those, can factor into those.

Moderator: Perfect. Moritz, if you want to go next.

von Knebel Moritz: Yeah, I’m gonna just add to that, more generally speaking, adding nuance to the dialogue. So overcoming the simple dichotomies of it’s innovation versus regulation. You can have innovation through regulation. We’ve had that for decades. Overcoming the US versus China or the human rights-based versus risk-based approach. Yeah, breaking away with these dichotomies that are not helpful and that ignore a lot of the nuance that is embedded within these debates.

Moderator: Moritz, I’m going to ask you to elaborate more because this is such an interesting point to make and adds so much value to the discussion. If you can elaborate more, I think we might give you one more minute.

von Knebel Moritz: Yeah, maybe I’ll touch on the, I mean, there’s much in here, but maybe the one thing I also touches on a question that was raised in the chat is on balancing ethics and innovation or regulation and innovation. And again, this is often pitted as a, you can choose. You can either regulate or you can innovate. And the EU is at a crossroads and we can’t fall behind. And yeah, we got to innovate, innovate, innovate. Whereas the reality on the ground is that safe and trustworthy frameworks that people, that give companies also predictability about the future and safety and know that their customers will like and feel that their products are safe is integral to innovation. And we’ve seen this in the past, the nuclear industry, the aviation industry, they took off after we had safety standards in place because that made it possible then to scale operations. And so I think this is just a very fascinating. for those of you interested, I’ve previously done work on this, writing up case studies on different safety standards in other industries and how they contribute to the development of a technology. So, yeah, going against the, we have to choose, the UK is an example of a very pro-innovative regulatory approach, and they didn’t step back and say, we’re not going to regulate. They were thinking about how regulation can serve innovation. And I think there are many ways that we can do this. And this often, unfortunately, gets ignored because it is easier to craft a narrative that pits two things and often two people and two camps against each other.

Moderator: Thank you so much. And what I take away from this is that building safe and trustworthy frameworks is due to these brilliant speakers. I also want to thank our audience, both here and online for their presence and for the conversation as well. Your presence on the final day of IGF 2025 is very important to us. And thank you to the wonderful IGF media team for coordinating the technical aspects. Before we close, I’d love to invite everyone on for a picture. First, we’re going to have a picture with the speakers, and then I’m going to invite the audience if they can come on the stage and have a picture with the speakers and myself. Thank you very much. Thank you. Thank you. The following is a work of fiction. Any resemblance to persons, living or dead, is coincidental and unintentional.

Agreement points

Need for multi-stakeholder approaches in AI governance

Speakers

– Alexandra Krastins Lopes
– Deloitte consultant

Arguments

Ethical foresight requires proactive mechanisms and organizational structures, not just theoretical principles

Multi-stakeholder approaches must include youth voices and create spaces for real co-creation

Summary

Both speakers emphasize that effective AI governance requires engagement across multiple stakeholders – Alexandra mentions engagement from product designers, data scientists, business leaders, and board-level commitment, while the Deloitte consultant specifically advocates for civil society, private sector, and public sector collaboration, with particular emphasis on including youth voices.

Topics

Legal and regulatory | Development

Frameworks must be adaptive rather than static

Speakers

– von Knebel Moritz
– Deloitte consultant

Arguments

Need for adaptive regulatory architectures with principle-based regulations and streamlined processes

Use case-based approach would be more effective than current risk-based categorization

Summary

Both speakers reject static, rigid regulatory approaches. Moritz advocates for adaptive regulatory architectures that can evolve with technology, while the Deloitte consultant proposes moving from risk-based to use case-based approaches that better account for context and application.

Topics

Legal and regulatory

International cooperation and coordination is essential

Speakers

– Alexandra Krastins Lopes
– Phumzile van Damme
– von Knebel Moritz

Arguments

Need for context-aware approaches that consider local realities rather than one-size-fits-all models

International law through UN processes could enable more inclusive AI governance

Need for shared language and consensus on key terms like systemic risk and AI definitions

Summary

All three speakers recognize the need for international coordination while respecting local contexts. Alexandra calls for global AI governance platforms with minimum safeguards, Phumzile advocates for binding international law through UN processes, and Moritz emphasizes the need for shared language and consensus on key terms.

Topics

Legal and regulatory

Current regulatory approaches are insufficient and reactive

Speakers

– von Knebel Moritz
– Alexandra Krastins Lopes

Arguments

Reactive frameworks that respond to known harms rather than anticipating emerging risks

Moving from ethical intention to institutional implementation within companies and organizations

Summary

Both speakers criticize current approaches as inadequate. Moritz identifies reactive frameworks as a key problem, while Alexandra emphasizes that ethical foresight cannot remain theoretical but must be institutionally implemented with concrete structures.

Topics

Legal and regulatory

Similar viewpoints

Both speakers advocate for flexible, principle-based regulatory approaches that avoid rigid definitions and false dichotomies. They both support the idea that regulation can enable rather than hinder innovation when properly designed.

Speakers

– Alexandra Krastins Lopes
– von Knebel Moritz

Arguments

Principle-based approaches without strict definitions allow technology and concepts to evolve

Adding nuance to dialogue and overcoming false dichotomies between innovation and regulation

Topics

Legal and regulatory | Economic

Both speakers emphasize the importance of understanding and building regulatory capacity, particularly recognizing the vast differences in resources and expertise across different countries and the need for capacity building initiatives.

Speakers

– Vance Lockton
– von Knebel Moritz

Arguments

Understanding different regulatory capacities across countries is critical for cooperation

Need for inclusive stakeholder engagement and capacity building across different sectors

Topics

Legal and regulatory | Development

Both speakers recognize that true democratization of AI requires more than just access to tools – it requires fundamental infrastructure development and participation across the entire AI lifecycle, with particular attention to inclusion of underrepresented voices.

Speakers

– Phumzile van Damme
– Deloitte consultant

Arguments

Need for democratization across entire AI lifecycle, not just access to tools

Infrastructure development is essential foundation for democratizing AI access

Topics

Development | Infrastructure

Unexpected consensus

Friction between tech companies and regulators is inevitable and potentially beneficial

Speakers

– Deloitte consultant
– Alexandra Krastins Lopes

Arguments

Critical thinking and assessment of AI outputs by citizens is essential for ethical AI use

Moving from ethical intention to institutional implementation within companies and organizations

Explanation

Unexpectedly, when asked about friction between tech companies and regulators, the Deloitte consultant stated that friction is inevitable and that democratic states ‘live from this friction,’ suggesting it’s a healthy part of the democratic process. Alexandra hoped to prevent such friction through multi-stakeholder discussions, but both acknowledged the reality of this tension while seeing potential positive outcomes.

Topics

Legal and regulatory | Economic

Youth inclusion is critical but often overlooked in AI governance

Speakers

– Deloitte consultant
– Phumzile van Damme

Arguments

Multi-stakeholder approaches must include youth voices and create spaces for real co-creation

Need for democratization across entire AI lifecycle, not just access to tools

Explanation

Both speakers unexpectedly converged on the critical importance of youth inclusion in AI governance. The Deloitte consultant specifically criticized multi-stakeholder approaches for forgetting youth voices, while Phumzile’s call for democratization across the AI lifecycle implicitly includes youth participation. This consensus was unexpected given their different professional backgrounds and regional perspectives.

Topics

Development | Sociocultural

Overall assessment

Summary

The speakers demonstrated strong consensus on several fundamental principles: the need for adaptive rather than static regulatory frameworks, the importance of international cooperation while respecting local contexts, the inadequacy of current reactive approaches, and the critical role of multi-stakeholder engagement including often-overlooked voices like youth.

Consensus level

High level of consensus on core principles with constructive disagreement on implementation approaches. This suggests a mature understanding of AI governance challenges and creates a solid foundation for collaborative policy development, though significant work remains in translating shared principles into practical, coordinated action across different jurisdictions and stakeholder groups.

Different viewpoints

Regulatory approach – Risk-based vs Use case-based frameworks

Speakers

– Deloitte consultant
– von Knebel Moritz

Arguments

Use case-based approach would be more effective than current risk-based categorization

Current AI regulation has more gaps than substance, with regulators lacking technical expertise

Summary

The Deloitte consultant advocates for moving from risk-based categorization to use case-based approaches, arguing that the same AI system can have different harms depending on application. Moritz focuses more on the fundamental gaps in current regulation and lack of technical expertise, suggesting the problem is more systemic than just the categorization method.

Topics

Legal and regulatory

Regulatory specificity – Principle-based vs Detailed frameworks

Speakers

– Alexandra Krastins Lopes
– Phumzile van Damme

Arguments

Principle-based approaches without strict definitions allow technology and concepts to evolve

Binding international law on AI platform governance is urgently needed

Summary

Alexandra advocates for flexible, principle-based approaches that avoid strict definitions to allow evolution, while Phumzile calls for binding international law, which would necessarily involve more specific and rigid legal frameworks.

Topics

Legal and regulatory

Innovation vs Regulation relationship

Speakers

– von Knebel Moritz
– Online moderator 2

Arguments

Adding nuance to dialogue and overcoming false dichotomies between innovation and regulation

Tech companies are pushing back against regulation they view as excessive and innovation-restricting

Summary

Moritz argues that the innovation vs regulation dichotomy is false and that regulation can enable innovation, while the online moderator highlights the real-world friction where tech companies view regulation as restrictive to innovation.

Topics

Legal and regulatory | Economic

Unexpected differences

Role of friction in democratic governance

Speakers

– Deloitte consultant
– Alexandra Krastins Lopes

Arguments

Critical thinking and assessment of AI outputs by citizens is essential for ethical AI use

Moving from ethical intention to institutional implementation within companies and organizations

Explanation

When asked about friction between tech companies and regulators, the Deloitte consultant stated that friction is expected and that ‘democratic states specifically live from this friction,’ while Alexandra hoped to avoid such friction through multi-stakeholder discussions. This represents an unexpected philosophical disagreement about whether regulatory friction is beneficial or problematic for democratic governance.

Topics

Legal and regulatory | Sociocultural

Overall assessment

Summary

The speakers showed remarkable consensus on high-level goals (ethical AI governance, international cooperation, inclusion) but significant disagreements on implementation approaches, regulatory specificity, and the role of friction in governance processes.

Disagreement level

Moderate disagreement with high convergence on principles but divergent implementation strategies. This suggests that while there is broad agreement on the need for ethical AI governance, the path forward remains contested, which could slow progress toward unified global AI governance frameworks. The disagreements reflect deeper tensions between flexibility vs. standardization, proactive vs. reactive approaches, and different regional perspectives on governance mechanisms.

Partial agreements

Similar viewpoints

Both speakers advocate for flexible, principle-based regulatory approaches that avoid rigid definitions and false dichotomies. They both support the idea that regulation can enable rather than hinder innovation when properly designed.

Speakers

– Alexandra Krastins Lopes
– von Knebel Moritz

Arguments

Principle-based approaches without strict definitions allow technology and concepts to evolve

Adding nuance to dialogue and overcoming false dichotomies between innovation and regulation

Topics

Legal and regulatory | Economic

Both speakers emphasize the importance of understanding and building regulatory capacity, particularly recognizing the vast differences in resources and expertise across different countries and the need for capacity building initiatives.

Speakers

– Vance Lockton
– von Knebel Moritz

Arguments

Understanding different regulatory capacities across countries is critical for cooperation

Need for inclusive stakeholder engagement and capacity building across different sectors

Topics

Legal and regulatory | Development

Both speakers recognize that true democratization of AI requires more than just access to tools – it requires fundamental infrastructure development and participation across the entire AI lifecycle, with particular attention to inclusion of underrepresented voices.

Speakers

– Phumzile van Damme
– Deloitte consultant

Arguments

Need for democratization across entire AI lifecycle, not just access to tools

Infrastructure development is essential foundation for democratizing AI access

Topics

Development | Infrastructure

Key takeaways

Ethical foresight in AI governance must move beyond theoretical principles to institutional implementation with proactive mechanisms, organizational structures, and board-level commitment

Current AI regulation has significant gaps with regulators lacking technical expertise and frameworks being reactive rather than anticipatory of emerging risks

International cooperation requires establishing shared language and consensus on key AI definitions, focusing on influencing design and development rather than enforcement

AI governance discussions lack global inclusion, particularly from the Global South, requiring democratization across the entire AI lifecycle rather than just access to tools

Effective AI frameworks should be use case-based rather than risk-based, implementable by developers, and include multi-stakeholder approaches with youth representation

Safe and trustworthy regulatory frameworks actually enable innovation rather than hinder it, as demonstrated in other industries like aviation and nuclear power

Citizens must develop critical thinking skills to assess AI outputs and question what they consume and create

Infrastructure development is essential for democratizing AI access globally

Resolutions and action items

Establish global AI governance platforms capable of harmonizing minimum principles on safety, ethics, and accountability while respecting local specificities

Create adaptive regulatory architectures with principle-based regulations and streamlined processes for technical implementation standards

Build independent technical advisory groups and capacity building programs for regulators

Develop international dialogues to establish consensus on key AI risk definitions and terminology

Create spaces for real co-creation between academia, public sector, private sector, and intergenerational voices

Implement binding international law on AI platform governance through UN processes

Shift narrative around AI from inevitability to focus on desired societal outcomes

Unresolved issues

How to achieve consensus on fundamental AI definitions and terminology across different countries and cultures

How to balance innovation-friendly approaches with necessary safety regulations without creating regulatory arbitrage

How to address the capacity and resource gaps between different countries’ regulatory authorities

How to effectively include Global South voices in AI governance when many countries lack basic AI policies

How to prevent AI fragmentation while respecting different cultural values and regulatory approaches

How to ensure meaningful youth inclusion in AI governance beyond tokenistic participation

How to implement cross-border enforcement mechanisms for globally operating AI platforms

Suggested compromises

Principle-based regulatory approaches that allow concepts and technology to evolve rather than strict definitions

Voluntary codes of conduct and soft law mechanisms that allow businesses to mitigate risks without restricting innovation

Risk-based approaches with flexible implementation adapted to specific local conditions and contexts

Sectoral regulatory approaches where existing authorities handle AI issues within their domains rather than creating new comprehensive AI laws

Regulatory sandboxes and iterative processes that allow testing of new approaches with quick feedback loops

Minimum safeguards with international coordination while respecting local specificities and strategic goals

I don’t think we have [well thought out systems]. I will employ a metaphor used by a former professor of mine who said that he does not have knowledge gaps because it would be just too large. But he has knowledge islands and those are surrounded by a lot of ocean that we don’t know about.

Speaker

von Knebel Moritz

Reason

This metaphor fundamentally reframes the entire premise of AI regulation discussion. Instead of assuming we have solid regulatory frameworks with mere gaps to fill, Moritz suggests we have isolated islands of knowledge in vast oceans of uncertainty. This challenges the conventional approach to regulatory development and highlights the magnitude of what we don’t know about AI governance.

Impact

This comment shifted the discussion from gap-filling to foundational questioning. It influenced subsequent speakers to think more holistically about regulatory frameworks rather than incremental improvements, and established a more humble, realistic tone about the current state of AI governance knowledge.

We need to establish global AI governance platforms capable of harmonizing minimum principles on safety, ethics, and accountability while respecting local specificities… international cooperation and legal certainty are essential for that to happen.

Speaker

Alexandra Krastins Lopes

Reason

This comment bridges the tension between global coordination and local autonomy – a central challenge in AI governance. Alexandra’s perspective from the Global South adds crucial nuance by acknowledging that one-size-fits-all approaches don’t work, while still advocating for minimum global standards.

Impact

This set up a recurring theme throughout the discussion about balancing international cooperation with local contexts. It influenced later speakers to consider how regulatory frameworks can be both globally coordinated and locally relevant, particularly affecting Phumzile’s call for international law and Vance’s discussion of regulatory cooperation.

There hasn’t been a proper democratization. While there may have been somewhat of a democratization in terms of access… we need to talk about it through the life cycle. So there needs to be democratization of AI development, democratization of AI profits, and democratization of AI governance itself.

Speaker

Phumzile van Damme

Reason

This comment deconstructs the popular narrative of AI ‘democratization’ and reveals it as incomplete. By breaking down democratization into development, profits, and governance phases, Phumzile exposes how current approaches only address surface-level access while ignoring deeper structural inequalities.

Impact

This reframing influenced the discussion to move beyond simple inclusion rhetoric to examine power structures more critically. It connected to Yasmin’s later emphasis on youth inclusion and reinforced the need for more fundamental changes in how AI governance is approached, not just who gets invited to the table.

We need frameworks that are actually implementable and understandable… if the developers cannot implement the regulations because they simply do not understand them, we need translators… but in an ideal world we would have developers who already understand this.

Speaker

Deloitte consultant (Yasmin)

Reason

This comment identifies a critical practical gap between regulatory intent and implementation reality. It highlights that even well-intentioned regulations fail if the people who must implement them cannot understand or operationalize them, pointing to a fundamental communication and education challenge.

Impact

This shifted the conversation from high-level policy design to practical implementation challenges. It influenced the discussion to consider not just what regulations should say, but how they can be made actionable by the technical communities that must implement them.

We need to shift the narrative around AI away from AI being either an inevitability or a wholly necessary piece of future economies, to think about what outcomes we want from future societies and how AI can build into those.

Speaker

Vance Lockton

Reason

This comment challenges the fundamental framing of AI discussions by questioning the assumption that AI development is inevitable or inherently necessary. It redirects focus from technology-driven to outcome-driven thinking, suggesting we should define desired societal outcomes first, then determine AI’s role.

Impact

This reframing influenced the final discussion segment and connected with Moritz’s closing comments about overcoming false dichotomies. It helped establish that AI governance should be purpose-driven rather than technology-driven, affecting how other speakers framed their concluding thoughts.

Overcoming the simple dichotomies of it’s innovation versus regulation. You can have innovation through regulation. We’ve had that for decades… safe and trustworthy frameworks… is integral to innovation… the nuclear industry, the aviation industry, they took off after we had safety standards in place.

Speaker

von Knebel Moritz

Reason

This comment dismantles one of the most persistent false narratives in tech policy – that regulation inherently stifles innovation. By providing concrete historical examples from other industries, Moritz demonstrates how safety standards actually enabled scaling and growth, offering a compelling counter-narrative.

Impact

This fundamentally challenged the framing used by many in the AI industry and provided a research-backed alternative perspective. It influenced the moderator to ask for elaboration and helped conclude the discussion on a note that regulation and innovation can be mutually reinforcing rather than opposing forces.

Overall assessment

These key comments collectively transformed the discussion from a conventional regulatory gap-analysis into a more fundamental examination of AI governance assumptions and power structures. The conversation evolved through several phases: Moritz’s ‘knowledge islands’ metaphor established intellectual humility about current understanding; Alexandra and Phumzile’s comments highlighted global power imbalances and the need for inclusive approaches; Yasmin’s focus on implementation practicality grounded the discussion in real-world constraints; and the final exchanges by Vance and Moritz reframed the entire innovation-regulation relationship. Together, these interventions created a more nuanced, critical, and globally-aware discussion that moved beyond surface-level policy tweaks to examine foundational assumptions about AI development, governance, and societal impact. The speakers built upon each other’s insights, creating a layered analysis that challenged dominant narratives while offering constructive alternatives.

How can we establish global AI governance platforms capable of harmonizing minimum principles on safety, ethics, and accountability while respecting local specificities?

Speaker

Alexandra Krastins Lopes

Explanation

This addresses the critical need for international coordination to prevent regulatory fragmentation while accommodating different cultural and economic contexts

What specific mechanisms can be developed to leverage AI for regulatory processes and governance?

Speaker

von Knebel Moritz

Explanation

Moritz mentioned that relatively little work has gone into using AI to improve regulatory frameworks, suggesting this as an underexplored area with potential

How can we develop adaptive regulatory architectures that can evolve quickly with technology rather than having static rules?

Speaker

von Knebel Moritz

Explanation

This addresses the fundamental challenge of regulatory frameworks being outpaced by rapid AI development

What would binding international law on AI platform governance look like in practice?

Speaker

Phumzile van Damme

Explanation

Van Damme identified this as her most urgent recommendation, but the practical implementation details need further exploration

How can we create effective spaces for real co-creation between different disciplines, sectors, and generations in AI governance?

Speaker

Deloitte consultant (Yasmin Alduri)

Explanation

While mentioning success in her work, the specific methodologies and scalable approaches for such co-creation spaces need further development

How can use case-based frameworks be practically implemented compared to current risk-based approaches?

Speaker

Deloitte consultant (Yasmin Alduri)

Explanation

This represents a fundamental shift in regulatory approach that requires detailed exploration of implementation mechanisms

What specific soft mechanisms can regulators use to influence AI system design and development rather than just enforcement?

Speaker

Vance Lockton

Explanation

Lockton emphasized this as critical but noted it’s a work in progress requiring further development of practical approaches

How can we systematically include youth voices in AI governance beyond tokenistic participation?

Speaker

Deloitte consultant (Yasmin Alduri)

Explanation

While identifying youth as having valuable solutions, the specific mechanisms for meaningful inclusion need further research

What are the specific case studies and methodologies for how safety standards in other industries (nuclear, aviation) contributed to innovation that can be applied to AI?

Speaker

von Knebel Moritz

Explanation

Moritz mentioned having done previous work on this but suggested more research is needed to apply these lessons to AI governance

How can we achieve democratization across the entire AI lifecycle – development, profits, and governance – not just access?

Speaker

Phumzile van Damme

Explanation

This represents a comprehensive framework for AI democratization that needs detailed exploration of implementation mechanisms

Summary

This panel discussion focused on how digital platforms, particularly TikTok, can be used to promote information integrity and combat misinformation. The session was moderated by Maurice Turner from TikTok’s public policy team and featured panelists from various organizations including a medical professional, representatives from humanitarian agencies, and advocacy groups. Dr. Ahmed Ezzat, a surgeon who creates medical content on TikTok, shared how he transitioned from casual content creation to serious public health communication after one of his posts about childhood infections reached 4.5 million views. He emphasized the importance of accountability and evidence-based practice for healthcare professionals on social media platforms.

Representatives from UNHCR, the Red Cross, and the Internet Society discussed their strategies for creating trustworthy content while facing significant challenges from misinformation campaigns. Gisela Lomax from UNHCR highlighted how misinformation directly harms refugee communities and can even contribute to forced displacement, citing the Myanmar crisis as an example. The panelists emphasized that effective content creation requires extensive expertise, collaboration across multiple disciplines, and significant time investment despite appearing simple to audiences. They stressed the importance of building trust over time rather than focusing solely on metrics like followers or views.

Key challenges identified included operating in polarized environments, competing against well-resourced misinformation campaigns, and balancing the need for quick responses with thorough fact-checking. The discussion concluded with recommendations for increased partnerships between humanitarian organizations, tech companies, and civil society groups to address information integrity challenges more effectively. The panelists agreed that while the battle against misinformation is complex and ongoing, strategic collaboration and maintaining high standards for content creation remain essential for success.

Keypoints

**Major Discussion Points:**

– **Platform strategies for promoting credible information**: Panelists shared how their organizations (TikTok, medical professionals, UNHCR, Red Cross, Internet Society) use digital platforms to disseminate trustworthy content, including partnerships with fact-checkers, creator networks, and evidence-based messaging to reach large audiences quickly and effectively.

– **Challenges in combating misinformation and disinformation**: Discussion covered the difficulties of fighting organized misinformation campaigns, including well-resourced disinformation “farms,” the polarized nature of online discourse especially in conflict zones, and the challenge of responding to false information once it spreads.

– **Building trust and credibility online**: Panelists emphasized the importance of accountability, transparency, using real names/credentials, engaging authentically with audiences, and building long-term relationships rather than focusing solely on metrics like views or followers.

– **Resource constraints and timing challenges**: A key tension emerged around balancing speed versus accuracy – the need to respond quickly to misinformation while taking time to verify information, assess risks, and coordinate with experts across multidisciplinary teams.

– **Collaborative approaches and partnerships**: Strong emphasis on the necessity of partnerships between humanitarian organizations, tech companies, civil society, academics, and content creators to effectively address information integrity challenges, especially given limited resources.

**Overall Purpose:**

The discussion aimed to explore how digital platforms, particularly TikTok, can be leveraged by credible organizations and creators to promote information integrity, share best practices for combating misinformation, and foster collaboration between different stakeholders in the information ecosystem.

**Overall Tone:**

The tone was professional yet conversational, with panelists sharing practical experiences and honest challenges. It remained consistently collaborative and solution-oriented throughout, with speakers building on each other’s points and emphasizing the shared nature of information integrity challenges. The discussion maintained an optimistic outlook about the potential for positive impact through digital platforms while acknowledging the serious difficulties involved.

Speakers

– **Maurice Turner**: Public policy team at TikTok, panel moderator

– **Dr. Ahmed Ezzat**: Surgeon in training doing general and breast oncoplastic surgery in London, creates clinical content and health news on TikTok

– **Eeva Moore**: Works at the Internet Society, focuses on educational content, advocacy, and community stories to help connect the remaining third of the world that’s not online

– **Gisella Lomax**: Leads UNHCR’s capacity on information integrity, works with the UN’s humanitarian agency protecting refugees, forcibly displaced persons, and asylum seekers

– **Ghaleb Cabbabe**: Marketing and advocacy manager of the IFRC (International Federation of Red Cross and Red Crescent Societies), based in Geneva

– **Audience**: Includes Bia Barbosa, a journalist from Brazil and civil society representative at the Brazilian Internet Steering Committee

Additional speakers:

None identified beyond those in the speakers names list.

# Digital Platforms and Information Integrity: A Multi-Stakeholder Perspective on Combating Misinformation

## Executive Summary

This panel discussion, moderated by Maurice Turner from TikTok’s public policy team, brought together representatives from healthcare, humanitarian organisations, and civil society to examine how digital platforms can promote information integrity and combat misinformation. The session featured Dr. Ahmed Ezzat, a surgeon creating medical content on TikTok; Gisella Lomax from UNHCR’s information integrity team; Ghalib Cabbabe from the International Federation of Red Cross and Red Crescent Societies; and Eeva Moore from the Internet Society. The discussion highlighted both the transformative potential of digital platforms for credible information dissemination and the complex challenges organisations face when combating well-resourced misinformation campaigns.

## Key Themes and Discussions

### Platform Reach and Impact

Maurice Turner opened by highlighting TikTok’s partnerships with fact-checking organisations and media literacy programmes designed to empower users. The panellists demonstrated the unprecedented reach that digital platforms offer for credible information dissemination. Dr. Ahmed Ezzat provided a compelling example, describing how one of his posts about childhood infections reached four and a half million views and was shared 156,000 times within 24 hours at zero cost, illustrating the massive public health potential of these platforms.

Gisella Lomax emphasised that UNHCR was “the first UN agency to create a TikTok account” and noted the vital importance of digital platforms for their work with displaced populations globally. She explained that these platforms are essential for providing lifesaving protection information. Ghalib Cabbabe outlined how the Red Cross utilises digital platforms for brand awareness, crisis communication, information sharing, and fundraising campaigns, citing recent examples including the Myanmar earthquake, Gaza, and Iran. Eeva Moore added that platforms serve a crucial role in connecting communities through educational content and stories.

### Content Creation and Trust Building

Dr. Ahmed Ezzat, who is part of the Clinical Creator Network in the UK, emphasised the importance of accountability for healthcare professionals, advocating for the use of real names and avoiding commercial endorsements that could damage credibility. He made a particularly insightful observation about public intelligence: “Actually, members of the public, regardless of what their understanding is, bearing in mind in the UK the reading age is about six or seven, the maths age is about three to four, they are phenomenally intuitive. They can sniff out right from wrong. You just need to be able to explain that information in an easy and simple way.”

Gisella Lomax outlined UNHCR’s approach to successful content creation, which requires creativity, partnerships with influencers, uplifting community voices, and making facts entertaining whilst remaining educational. She stressed the importance of building partnerships with tech companies, academic institutions, and civil society organisations.

Eeva Moore provided crucial insight into the hidden complexity of content creation, noting that expertise must be “baked into” the production process with multiple layers of review. She explained that whilst content should appear simple to audiences, it requires extensive behind-the-scenes work. Ghalib Cabbabe reinforced this point, emphasising that organisations must rely on field experts rather than just marketing teams and prepare proactive strategies including scenario planning.

### The Challenge of Misinformation

A central theme emerged around the asymmetric nature of the battle against misinformation. Dr. Ahmed Ezzat described how misinformation creates an unfair defensive position where evidence-based voices must defend against accusations whilst being constrained by factual accuracy.

Gisella Lomax connected online misinformation to devastating real-world consequences: “Information risks such as hate speech and misinformation are directly and indirectly causing real world harm. And I mean violence, killings, persecution. It can even be a factor in forced displacement, in causing refugees… as we saw in Myanmar back in 2016, 2017, when hate speech… had a decisive role in the displacement, I think, of 700,000 Rohingya refugees into Bangladesh who are still there today.”

Ghalib Cabbabe highlighted the resource disparity, noting that organisations face opponents with significant resources including state actors and misinformation farms operating in highly polarised environments.

### Time as the Fundamental Challenge

Eeva Moore identified time as the core challenge: “You could boil the challenge down perhaps to one word, and that’s time. If you’re in the business of trying to disrupt or break something, you get to move at a much faster pace than if you’re on the other side of that equation.”

This time constraint creates tension between speed and accuracy. Ghalib Cabbabe advocated for balancing speed with thoroughness by assessing risks before communicating, sometimes taking time to understand how misinformation evolves rather than rushing responses. Eeva Moore emphasised the need for creative solutions and pre-prepared resources to overcome these constraints.

### Platform Responsibility and Resource Concerns

Gisella Lomax raised concerns about platform responsibility, particularly regarding content moderation in less common languages: “We have seen, to our dismay, a weakening of these capacities and perhaps less resourcing from some companies. And I would extend that, for example, to content moderation in less common languages… Are you adequately providing content moderation in less common languages in these very volatile contexts where you don’t actually have a business argument?”

This exposed a critical gap in platform safety measures for vulnerable populations who communicate in languages that aren’t commercially viable for platforms.

## Areas of Consensus and Strategic Differences

The panellists agreed that digital platforms provide unprecedented reach for information dissemination and that content creation requires multi-layered expertise across organisations. All speakers recognised that misinformation poses serious real-world threats requiring proactive, systematic responses.

However, some tactical differences emerged around response strategies. Dr. Ahmed Ezzat emphasised the defensive disadvantage of being constrained by evidence, whilst Ghalib Cabbabe advocated for strategic patience. Eeva Moore focused on speed through creative solutions and pre-preparation.

## Audience Engagement and Future Collaboration

The discussion included audience participation, with Brazilian journalist Bia Barbosa asking questions about platform engagement. Gisella Lomax actively invited collaboration from academics, tech companies, and NGOs, specifically mentioning a Wednesday 2 PM event showcasing UNHCR’s South Africa pre-bunking project and referencing “Katie in the pink jacket” for those interested in UNHCR’s information integrity toolkit.

The panellists recommended examining comment sections on Red Cross digital platforms to understand the complexity of misinformation challenges firsthand.

## Key Recommendations

Strategic recommendations included:

– Focusing on quality of engagement rather than vanity metrics

– Building partnerships to share resource burdens

– Preparing proactive content strategies rather than only reactive responses

– Collaborating across organisations rather than building all capabilities internally

– Developing scenario planning and pre-prepared responses

## Unresolved Challenges

Several challenges remained unresolved, including how to adequately resource content moderation in less common languages, addressing weakening trust and safety capacities from some platforms, and overcoming the fundamental time asymmetry between misinformation creators and fact-based responders.

## Conclusion

Maurice Turner’s closing remarks emphasised the tensions revealed in the discussion and the key takeaways about collaboration and resource sharing. The panel demonstrated that effective information integrity work requires sophisticated understanding of platform dynamics, audience psychology, and collaborative resource management.

As Eeva Moore noted, this work “should be labour-intensive” and should “look like a light lift, but actually, in fact, be a pretty heavy one.” The discussion revealed that information integrity is not merely a technical challenge but a humanitarian imperative with life-and-death consequences, requiring sustained collaboration between humanitarian organisations, tech companies, civil society, and academic institutions.

Maurice Turner: Welcome to the folks in the audience here in person as well as online. We’re going to be discussing using digital platforms to promote information integrity. My name is Maurice Turner and I’m on the public policy team at TikTok. Hello, hello. Welcome everyone who’s here in person as well as online. Welcome everyone who is here in person and online. My name is Maurice Turner. My name is Maurice Turner and we’re going to be discussing using digital platforms to promote information integrity. Again, I want to welcome everyone here in person and online. And for our panel discussion, we’re going to be talking about using digital platforms to promote information integrity. I have several panelists here and I’d like to start off with a brief introduction to how we’re facing this challenge at TikTok. And then I will get into introductions for our panelists. And we will jump right in to a back and forth that will also include Q&A from the audience. So if you have questions, please do make a note of them during this panel discussion. And towards the end, we will have microphones on both sides of the table so that you can get up and ask your questions. I’m also encouraging questions from our audience online. So if you have any questions, feel free to go ahead and type those in. And we have someone monitoring that online discussion. So your question can be asked toward the end of the session. At TikTok, we believe this conversation is important because information integrity itself is important, both to inspire our creators to promote credible information, but also to ensure that organizations are doing important work to amplify their own messages. We look forward to hearing more about organizations doing that work in our discussion later on today. At TikTok, we remain committed to using fact-checking to make sure that the content on our platform is free from misinformation. We partner with more than 20 accredited fact-checking organizations across 60 different markets. We do this work continuously to ensure that we are making sure that our platform is as free from misinformation as possible. And we also empower our users through media literacy programs to give them resources to recognize misinformation, assess content critically, and report any violative content. And now for our panelists. I’d like to go ahead and start off with introductions. Dr. Ahmed, he’s a creator that produces content. We also have Eva from the Information Society, as well as Gisela from the UNHCR. As an icebreaker, let’s go ahead and start off with the introduction. Dr. Ahmed, how would you like to introduce yourself and talk about how you produce content on TikTok?

Dr. Ahmed Ezzat: Thank you very much for the introduction. So I’m Dr. Ahmed Izzat. I’m a surgeon in training doing general and breast oncoplastic surgery in London. I create clinical content and my focus is health news. And I also do lots of content strategy and campaigns. And one of the draws to being on a platform like TikTok is the force of good it can portray in the way that it’s quite sensationalist, that if you get the balance right from a public health perspective, for instance, in the UK, there’s been a heat wave. You can reach a million people, half a million people within less than 24 hours at zero cost, essentially, which is a massive gain, I find. So I think there’s a massive power of good if it’s harnessed well using evidence-based information.

Maurice Turner: And Eva, how do you use the platform?

Eeva Moore: At the Internet Society, we are really focused on a couple types of content. Educational. We work to help connect the remaining third of the world that’s not online. And that can look like building infrastructure in one of the 181 countries where we have chapters. So a lot of it is educational. There’s a lot of advocacy. And then there’s a lot of community stories. How do we demonstrate the impact that we’re having in the world without making ourselves the center of the story itself? Thanks.

Gisella Lomax: Hi, everyone. My name is Gisela Lomax, and I lead UNHCR’s capacity on information integrity. For anyone not familiar with UNHCR, and you might not be familiar because certainly this is my first time at the IGF, we’re one of the UN’s largest humanitarian agencies tasked with protecting the world’s refugees, forcibly displaced, asylum seekers. That’s 123 million people now across, I think, 133 countries. So we’re using social media in many sophisticated and extensive ways. It’s vital. As Dr. Ahmed said, the reach is massive. One, to provide vital, lifesaving protection information to communities, both in emergency context, protracted context. And the other is to inform the general public as well as specific audiences and stakeholders, including government partners, civil society, the private sector, about our work. To lift up the voices of our refugee partners as well, and amplify communities, and to inspire. And I would say as a kind of a little unknown fact is that we were the first UN agency to create a TikTok account quite a few years ago now. And I think it’s very interesting the way TikTok really seeks to make important information fun and entertaining. Although that also comes with challenges, but I think we’ll come to that.

Maurice Turner: We’re also joined online by our panelist, Ghalib from the Red Cross. Would you please go ahead and introduce yourself and let us know how you are using TikTok?

Ghaleb Cabbabe: Sure. First, apologies for making this panel hybrid again. And it’s nice to connect even online. And hi, everyone. So I’m Ghalib. I’m the marketing and advocacy manager of the IFRC, the International Federation of Red Cross and Red Crescent Societies, based in Geneva. How do we use digital platforms? Because we’re also tackling different fields, different topics. So depending on the situation, it would be, let’s say, different platforms. But usually the main use is, of course, for brand awareness to make sure that people know what we do, how we do it. We also use digital platforms. And this is a big part of the work in crisis mode sometimes when, for example, we have a disaster, an earthquake, let’s say, lately in Myanmar or the situation also in Gaza or Iran. So this is more in crisis mode. Sometimes it’s to share information. Sometimes it’s to try not to have misinformation. I think we’ll open also this topic more in detail in the next coming minutes. And sometimes we also use digital platforms for direct or indirect fundraising campaigns. So we’re happy to be on TikTok. And I can recall working with the Red Cross that was maybe back in 2017, where it was, we can say, the beginning of the platform. And to see where it is today, I think that’s quite a great achievement. Thank you.

Maurice Turner: I’m going to go ahead and start us off with a specific question for Dr. Ahmed. How did you get your start on TikTok? And what are some of the ways that you achieve your goal of making sure that you are getting content out on the platform and out to an audience specifically related to STEM inequality?

Dr. Ahmed Ezzat: So I think I’ll share my journey because I think it’s representative and it’s an honest one. So when I first started my journey on TikTok as a medical professional and academic surgeon, I wanted to set up my content for actually very different reasons, which is trying to do stuff that’s seemingly for fun, lighthearted. But then I absolutely did not want to be that cliche medic who’s creating content on medicine. But then it was too difficult to resist because there was such a need. And we are quite fortunate in the UK. There’s a very nice and well-organizing climate of clinicians that create content. And then I had my first bit of content, which went viral and it hit four and a half million views. And this was on infections in childhood. At the time, there was an outbreak. And then this was shared 156 times. 6,000 times. And I thought, well, the UK population is, you know, in the in the in the 60 plus million four and a half million is phenomenal. So from that point on, I actually really started to look into tick tock in a very different way, because the impact is so phenomenal. So that if there were outbreaks on say, food poisoning, I’ve had agencies give me information, but say, well, we can’t disclose it publicly just to try and govern misinformation to try and help information. And so I shifted from a name that I had to begin with, which was a nickname, to actually moving on to using my real name on tick tock, which is to celebrate the whole and leverage the whole point of the fact that you should be held accountable for the medical the medical information that you say. And I really do also see a massive shift in healthcare professionals and colleagues of mine, who to begin with, used to look at social media, tick tock, you know, as a as a as a as a vanity project. But then now, government have been in, you know, been working with us, you know, institutions that are verified, you know, political organizations to charities to esteemed companies, because they see the value and the reach that you can have just by creating this content. But the really important bit, which is going to your point, how do you go about achieving your goals? tick tock, for example, has been fantastic in trying to balance freedom of information against integrity of information. And so we work together to set up the clinical creator network in the United Kingdom. And it’s through having some, you know, microclimate of clinicians that are really there to do a force of good, but also using evidence based practice in the same way that they would be accountable to the General Medical Council or whichever organization they’re at wherever in the world. I think that’s the most important thing. Because as a healthcare professional, you’re not a lifestyle creator, meaning that if I was to buy a brand of jeans tomorrow and put it on my posts, no one would care. But if I was to suddenly pick up a brand deal to try and promote weight loss medicines, for example, which is an allure that we get every day, then this would absolutely decimate credibility. And so you really have to carry the responsibility extremely carefully as a clinic, as a clinician.

Maurice Turner: Now, a question that I get pretty regularly is how do I make content that’s popular on the platform? And I’m not the expert in that. So I’ll leave it to others. But I think a related question is, what are some of the strategies that might be used to push out content that is actually trustworthy and credible, so that people on the platform are getting the information that they’re looking for? Gisela, do you have any sort of a response for that?

Gisella Lomax: Yeah, definitely. Well, I was going to add to how we use our strategies for social media platforms and digital channels. And I think first on the communication side, that’s very much being creative. It’s partnering with influencers and brands, uplifting refugee-led storytelling, using the facts, but trying to put them across in an entertaining, educative way. So all of this good stuff to amplify refugee voices, spark empathy, drive action offline, as well as online. However, the challenge, of course, is this is increasingly undermined and threatened by the growth of misinformation, disinformation, hate speech, these types of information risks. And so my role, and I used to work more on the communication side, and now I pay tribute to our communications colleagues and the social media team at UNHCR and at countries who do a fantastic job. My work now is on addressing these risks. So at UNHCR, we have this information integrity capacity, and then field-based projects around the world in around nine countries, basically developing a humanitarian response to mis- disinformation and hate speech. We recently launched a toolkit. If you Google UNHCR information integrity, you’ll find it, which has got practical tools, guidance, common standards on all of this quite technical stuff, such as assessing the problem and then different types of responses. And we really have tested a plethora of responses, and I can highlight just a couple, given the time. Obviously, there’s continuing to fill the feeds with proactive, trusted, reliable, credible, accessible, multilingual, the list goes on, content and information. And working with technical, with digital platforms, including TikTok, to try and get this content uplifted, because it often naturally isn’t the sort of content that rises to the top of algorithms. And so those partnerships are key. But the other aspects are how do we deal with a spread of, say, hate speech that’s inciting violence, disinformation that’s undermining humanitarian action, misinformation that’s making people feel, you know, more hostile or mistrustful of communities. And I’d like to highlight one quite exciting response. And if I may, a little plug, we have a really interesting project in South Africa, testing the notion of pre-bunking. This is the inoculation theory of trying to help societies and communities become more resilient to hate and lies in partnership with Norway, South Africa. But we’ve also benefited from some technical support from TikTok and from Google and from others as well. And we have an event where we’ll be showcasing this on Wednesday at two o’clock, and my colleague Katie there in the pink jacket can tell you more. So that’s a really positive example. But as you can see, it really runs the gamut, and I’m happy to go into a bit more detail on some of those.

Maurice Turner: Ghaleb, I’ll pass it over to you. As marketing manager, what are some of the strategies that you use to get out that trustworthy information?

Ghaleb Cabbabe: It’s a very, let’s say, timely question. It’s an important one, not an easy one, because unfortunately, we don’t have today all the tools. Because if we look today at the environment of misinformation, let’s face it, we have sometimes companies, farms in some places where their only goal is to produce misinformation. And usually, it’s very hard to fight against these in terms of resources, in terms of means, in terms of targets. So what we do, of course, internally, is always check, double check, and check again, the information we want to share, the information sometimes we want to react on, we want to comment on. I’ll give you an example. It’s been unfortunately the case in the past few weeks and months with the killing of PRCS, Palestinian Red Crescent colleagues in Gaza, for example, in Iran lately as well, where we had the Iranian Red Crescent colleagues killed in the latest attacks. So first, checking the information, rechecking at local, at regional level is, of course, something that has to be done. We also rely on the experts in their own fields. Of course, we’re the marketing, we’re the social media team. But it’s not us who make sometimes the call to a certain line, to a certain reactive line or key messages that we want to share. So relying on experts is really essential. And also the monitoring part, making sure that we try as much as possible to anticipate what could go wrong, to anticipate information that’s going out, and that could lead to a misinformation situation. So the monitoring part is also important. And also, we are always trying to be, trying to be again, because it’s not an easy battle, a step ahead in terms of being proactive. This would be, for example, by preparing reactive lines, by also trying to identify the different potential scenarios and be ready if and when it happens. So these are strategies that we put in place. Again, it’s not an easy file. It’s a very essential one. And we’re putting resources, we’re also as much as possible trying to train staff internally, to avoid situations like this, because sometimes also, this type of misinformation is triggered by a certain, let’s say, tweet or information or post, be it on TikTok or other platforms that could also trigger a reaction, a comment that can cascade. So being proactive, training also staff is part of the different strategies. And let’s not forget that the main objective of our communication is, of course, to be as impactful as possible. And we see this information as a threat, not only to the communication that we share on social media, but also sometimes to the credibility of the organization.

Maurice Turner: Thank you. Those were quite interesting strategies. Eva, what about you?

Eeva Moore: Kind of building on that, I think I think you have to bake expertise into the production process. Anything that you see on the social media platforms should be easily digestible, but it should be built on top of multiple layers of expertise and conversations and trust that we have across our community. We do advocacy efforts in countries that I may never have visited, which means I’m not the person to create that. And so it should be labor-intensive. I mean, dealing with these types of issues when we’re creating them, when we want to be accurate, should look like a light lift, but actually, in fact, be a pretty heavy one. And that takes, I think, just incredible amounts of conversation and listening and being very online and seeing what others are saying. I mean, you have to be consuming a lot of information. Sadly, that includes coming across the disinformation, if you’re going to understand it and to understand how it’s navigating the space. And the expertise also has to be credible, because your audience doesn’t want to be spoken down to. So when I say expertise, that is legal, that is technical, but it’s also the people who are impacted by the – I mean, you touched upon refugee communities. For us, it’s about the credibility of people. We can all be led down a disinformation path, but I think people have a pretty keen sense of whether or not they want to listen to somebody, and credibility is built into that. So I think relationships, it’s sort of the world that you inhabit in real life, how it manifests online is, I think, reflected into that work. So you just have to bake it in, and you have to have your work reviewed, and you have to be willing to produce things that maybe don’t make it out into the world. You also have to be just willing to see – to do your best and listen to others and make sure that by the time it gets out there, it’s accurate.

Maurice Turner: I think that’s so key in that it takes the consumption of so much information and the recognition that not one single person or even one organization has all the expertise. There’s a reliance on other folks in the community in building those relationships so that you can get that information, distill it, and at the end of the day put out a product that looks like it was easy to make because it was actually easy to understand, and there’s quite a bit of work that goes into that. I’m sure all of us, not only on stage and online as well as in the audience, have come across challenges in information integrity. So I’d like to shift the conversation to hear more about some of those challenges that we face specifically, and then also how we were able to overcome those challenges. Dr. Oppenman, can you share a challenge that you faced and how you were able to overcome that in terms of information integrity and maybe even an example where you weren’t necessarily able to overcome it?

Dr. Ahmed Ezzat: Yeah, absolutely, and some fantastic points here. I think just to reiterate, one of the strong points of social media, especially short-form social media, is that it can really reach those with lived experiences, and you are very much there, very much accessible, and very much accountable to the hundreds of comments you get. But of course, disinformation sometimes is difficult to combat. Once it’s out, it’s very difficult, but you’re almost on the defense, almost trying to defend an accusation, and sometimes it is an unfair footing because you can’t go outside of descriptive evidence. But I remember when I, and this was working after a call by a Labour MP, who now they’re in government, who had called out for a, oh, why wouldn’t it be perfect to do a campaign on the MMR vaccine, because there was a rise, and TikTok responded and said, oh, well, we’ll do it. And then I fronted that campaign, and I remember we filmed up in the Midlands of England, in London, in different demographics of different ethnicities, backgrounds, etc. And I actually remember the disconnect between the reality, which was that many people were believing in the vaccination importance, but then this was after the backdrop of the COVID vaccination saga, where first of all, there was a time where these medicines were tested in Africa. That created catastrophes in terms of bringing back lots of bad memories around it, historical memories. But then you had lots of top-down suited and booted officials telling people what to do, and that made things even worse. And the result was that the public health message has transformed forever, in that people really don’t want to be expected to listen to officials telling them what to do. And this is where social media had risen. But then I received a massive backlash when I did the content on the MMR vaccine, and with very, very, very loud voices who are a very, very, very small minority spreading disinformation. And I think going back to what you’ve said, that if you build trust and people know that you are doing your best and understand your values over a period of time, as opposed to just a one-off hit, then that tone starts to soften. And the other thing I would just say is, and I’ve had this conversation with very senior officials from another part of the world, who said, well, you may say that your group of viewers are intuitive, but ours aren’t. And this is another well-developed nation, one of the most well-developed and one of the richest. And I was actually saying it’s exactly this mindset which sets us back. Actually, members of the public, regardless of what their understanding is, bearing in mind in the UK the reading age is about six or seven, the maths age is about three to four, they are phenomenally intuitive. They can sniff out right from wrong. You just need to be able to explain that information in an easy and simple way. And the last thing I’d say, so I don’t take up too much time, when you’re looking at content quality, please don’t look at followership. Don’t look at likes. Look at engagement and quality of engagement in the comments and the type of content. Because if the content is high quality, it will deliver on engagement. I wouldn’t worry too much about the followerships and the millions of views. It’s about the quality of engagement.

Maurice Turner: Ghaleb, can you share a particular challenge that you’ve faced and maybe offer some recommendations to the audience about how you were able to overcome them in this particular space?

Ghaleb Cabbabe: Sure, actually. To give us maybe an indication on the situation that we’re dealing with and that sometimes we have to face. I believe and maybe I invite also people in the audience to maybe go on our digital platforms and also the digital platforms of maybe also other Red Cross and Red Crescent movement partners like the ICRC and check the comments section. And this is where you would see how complex this environment is. Because today, unfortunately, we are moving, working, reacting, engaging in a very, very polarized world. And this is even more present because we are talking and our work is in war zones, conflict zones. So this is even more polarized. So misinformation is even more present here. So I think the challenge here is really the nature of what we’re facing when we talk about misinformation. As I was saying previously, there are people whose work is only to create this type of misinformation and to feed it and to make it grow online. So sometimes the battle is not an easy one because the opponent, if we can call it so, of misinformation is an opponent with sometimes many resources. You have sometimes states behind it. So I think the challenge comes from here, the very polarized context and also the people who could be behind this misinformation. How we’re trying to deal with this, again, is by sharing, cross-checking, and checking again the information before sharing it, making sure sometimes also in very sensitive context to anticipate what could be the risk. This is really very important to assess the risk before communicating. And although we are in a world on digital platforms, not only TikTok, where you have to be prompt, you have to be really fast in terms of reaction, one of my recommendations would be that sometimes it’s better not to rush. You have to be prompt, of course. You have to be fast, reactive. But sometimes by taking the time to assess, to see how maybe a certain piece of misinformation is evolving online, the direction it’s taking, it gives you maybe a better understanding of the situation and a better way to tackle it. So, yes, being prompt and fast, but do not rush. It’s a daily work. It’s an ongoing work. It starts, actually, before the misinformation it shares. It starts with what we plan to produce, what we plan to communicate on in our statements, in our press release, in our interviews with the press. And again, it was mentioned in the panel, it’s not only a social media team work. It’s the work of the whole team in terms of expert on the legal side, on the media side, on the different expertise. So it’s an ongoing work, a daily work. And again, it’s a complex one, and it’s not an easy battle.

Maurice Turner: Thank you. And it seems that not only doing the work beforehand to prepare and have a process, but also having the confidence and the patience to be able to understand what you’re going to be responding to is part of the strategy in responding to that challenge. Gisela, do you have a particular challenge related to information integrity that you’d be able to share with us and maybe some recommendations for how you’re able to respond to that?

Gisella Lomax: Thank you. And I like you say a particular challenge. There are many, but perhaps I could just highlight three, but then there are some recommendations I can make as well. I think building off the points thatGhaleb Cabbabe just shared, the first is to say this is… Sorry, this is a very jargonistic term. This is multifunctional work. This is not just communications. This is protection, policy, operational, governance and communications. And so really it’s getting that message across and then making sure these multifunctional teams are resourced. That’s more of an internal challenge for how we address it. But I would say, I mean, there’s three very distinct challenges. And one is that protection challenge. Information risks such as hate speech and misinformation are directly and indirectly causing real world harm. And I mean violence, killings, persecution. It can even be a factor in forced displacement, in causing refugees. In fact, as we saw in Myanmar back in 2016, 2017, when hate speech, which had already been circulating for some time, really exploded and had a decisive role in the displacement, I think, of 700,000 Rohingya refugees into Bangladesh who are still there today. And then also on communities, it’s normalising hostility towards refugees and migrants. So the problem, the harms over time, the very direct harms, that’s on the community, the refugee side, on the operational side. It’s increasingly hampering humanitarians from doing their job around the world in conflicts in Sudan, for example, or other places, disinformation narratives about humanitarians perhaps politicising their work can also erode people’s trust. And, you know, trust is everything. And we know that trust can take years or decades to build and can be destroyed very, very quickly. And I think I recognise the importance of individual voices and I think I celebrate what you’re doing, Dr Ahmed, and people like you, but let’s not forget that we also need to help institutions remain trusted as well. If people can’t trust public health bodies or UN entities, then we also have a challenge. And then a third, very more specific one, is that of trust and safety capacities from digital platforms and tech companies. We have seen, to our dismay, a weakening of these capacities and perhaps less resourcing from some companies. And I would extend that, for example, to content moderation in less common languages. You know, I think this is always a question I ask to all tech companies. Are you adequately providing content moderation in less common languages in these very volatile contexts where you don’t actually have a business argument? You’re not selling ads or making money, but your platforms are still delivering fundamental information. So there are three challenges. But on to a recommendation. I think we’re sat here at an event designed to create international cooperation to solve these problems, so it’s partnerships. And I think speaking as a humanitarian, for us, it’s increasingly building partnerships with tech, with digital, digital rights organisations, civil society NGOs that bring in the skillset, expertise and knowledge that we don’t have, and given a very dramatic and grave financial situation for the humanitarian sector, we’re not going to have more capacity. So how can we partner, collaborate to bring in that knowledge, to test these responses, innovative, long over time, like digital literacy, more immediate for these kind of quite acute harms and conflicts? So my recommendation is to please talk to us. If you’re academic, there are many research gaps and we can help give you access to try and fill those. Then we can take that research to inform strategies and policies. If you’re a tech company, I’ve already mentioned some of the ways that we can help each other. If you’re a digital rights, an NGO, I think there’s many more. So that’s both a recommendation and a plea from us, and that’s what brings me here to Oslo this week.

Maurice Turner: Eva, I’d like to hear more about maybe more of the technical side of the challenges. What are some of the challenges that you’re hearing about from the folks that you’re working with and how they’re trying to tackle information integrity, and what are some of those recommendations that seem to have surfaced to the top?

Eeva Moore: You could boil the challenge down perhaps to one word, and that’s time. If you’re in the business of trying to disrupt or break something, you get to move at a much faster pace than if you’re on the other side of that equation. So from a purely practical standpoint, that means, as comms people, leaning into the creative side of it. So how can we tell a compelling story on a timeline that mirrors, say, a news cycle? We did that with the proposed TikTok ban in the United States. That was a timeline for which this long period of review did not lend itself well to. But we have the experts. They had already done the work. So my very qualified, fantastic colleague Celia went in and just plucked headlines from a blog post, designed it within seconds, and put it out there. Not seconds, perhaps minutes. So it’s knowing what resources are already there that you can act on, because I agree with what Gallup said, is that sometimes you have to wait, but there’s also the aspect of it where sometimes you can’t afford to, right? So the creative side of it, which is in and of itself interactive, but that’s one way around time. The other way is simply doing the heavy lifting for your community, remembering that if you’re the person whose job it is to create these things, you are not asking somebody. We work with volunteers, essentially, around the world, right? These are people with lives, with families, with jobs, often in challenging circumstances as well as less challenging circumstances. It’s my job, it’s my team’s job, to try to ask as little of them so that then when we do ask, they’re able to actually jump in and help. But time, it’s labor intensive. It’s not something that people are swimming in.

Maurice Turner: It seems like that’s a poignant point to wrap up that part of discussion, time. It’s a resource that no one quite has enough of, and as we heard earlier, there needs to be a balance of when do you find that right time do you do it fast, do you wait to be more thorough and what’s appropriate for the response that you need given that particular situation. At this time, I’d like to go ahead and open up the time for questions from the audience. Again, we have microphones, so if you have a question, feel free to step up to the microphones on the side. If you’re a little bit shy, that’s okay too. You can just raise your hand, maybe pass a note to someone who might have a question, and if you’re online, please do type in a question there and we’ll have that moderated as well.

Audience: May I? Hi, good afternoon. Thank you for the panel. My name is Bia Barbosa. I’m a journalist from Brazil. I’m also a civil society representative at the Brazilian Internet Steering Committee. Thank you for all the sharing of your experience producing content for TikTok, specifically if I got it, or for social media in general. And how you face these challenges regarding information integrity for us as a journalist perspective is something that is in our most… This is one of the most important preoccupations at this time. And I would like to hear from you because regarding the challenges, I didn’t see, besides Gisela mentioning, the resources. Thank you very much.

Maurice Turner: Do we have any other questions from the audience? Feel free to step up to the microphone if you have one. Any questions from the online audience? Excellent. Well, what I’ll go ahead and do is wrap us up. Feel free to engage with our panelists as we exit the stage and close out the session, or if not now, then go ahead and find them throughout the week. I think we had an interesting discussion in hearing about how these organizations that you all represent utilize the platform or utilize digital platforms to put out information to push back against some of that misinformation that’s out there. We also heard about some strategies for attacking the safety issue regarding information integrity, and we also heard some of the recommendations that came through as well. I particularly enjoyed hearing about those tensions that naturally take place between having to be viewed as experts, recognizing that you also rely on a network of expertise, so it’s not just one individual or one sole organization. And also that tension between balancing that limited resource of time, so being very prompt, but also allowing time for that situation to evolve to better understand how best to respond to it in a way that is not only authentic, but in a way that is also prompt, so that even when attentions might be shortened, that message can get out effectively. So again, please do join me in thanking the panel, and I’d also like to thank you all for joining this discussion, and I hope you enjoy the rest of your week at the conference. Thank you. Thank you.

Agreement points

Digital platforms provide unprecedented reach and impact for information dissemination

Speakers

– Maurice Turner
– Dr. Ahmed Ezzat
– Gisella Lomax
– Ghaleb Cabbabe
– Eeva Moore

Arguments

TikTok partners with over 20 fact-checking organizations across 60 markets and empowers users through media literacy programs

Medical professionals can reach millions of people within 24 hours at zero cost using evidence-based information, representing massive public health potential

Digital platforms are vital for providing lifesaving protection information to 123 million displaced people across 133 countries

Digital platforms serve multiple purposes including brand awareness, crisis communication, information sharing, and fundraising campaigns

Platforms help connect the remaining third of the world that’s not online through educational content and community stories

Summary

All speakers agreed that digital platforms, particularly TikTok, offer massive reach and potential for positive impact when used strategically for information dissemination, whether for health information, humanitarian aid, or global connectivity.

Topics

Sociocultural | Development | Human rights

Content creation requires multi-layered expertise and cannot be done by single individuals or departments alone

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax
– Ghaleb Cabbabe
– Eeva Moore

Arguments

Healthcare professionals must maintain accountability and use real names to build credibility, avoiding commercial endorsements that could damage trust

Build partnerships with tech companies, academic institutions, and civil society organizations to leverage expertise and resources that humanitarian organizations lack

Organizations must rely on field experts rather than just marketing teams, and prepare proactive strategies including reactive lines and scenario planning

Expertise must be baked into the production process with multiple layers of review, making content appear simple while requiring heavy behind-the-scenes work

Summary

All speakers emphasized that effective content creation requires collaboration across multiple expertise areas, from medical professionals to legal experts to field specialists, rather than relying on single individuals or departments.

Topics

Sociocultural | Legal and regulatory | Human rights

Misinformation poses serious real-world threats that require proactive, systematic responses

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax
– Ghaleb Cabbabe

Arguments

Misinformation creates an unfair defensive position where evidence-based voices must defend against accusations while being constrained by factual accuracy

Information risks directly cause real-world harm including violence, killings, and forced displacement, as seen with Rohingya refugees in Myanmar

Organizations face opponents with significant resources including state actors and misinformation farms operating in highly polarized environments

Summary

Speakers agreed that misinformation is not just a communications challenge but poses direct threats to safety, security, and organizational credibility, requiring systematic and proactive responses.

Topics

Human rights | Cybersecurity | Legal and regulatory

Time constraints create fundamental challenges in responding to misinformation effectively

Speakers

– Ghaleb Cabbabe
– Eeva Moore

Arguments

Balance speed with thoroughness by assessing risks before communicating, sometimes taking time to understand how misinformation evolves rather than rushing responses

Time constraints create fundamental challenges since disruption moves faster than constructive response, requiring creative solutions and pre-prepared resources

Summary

Both speakers identified time as a critical constraint, noting that while misinformation spreads quickly, effective responses require careful preparation and strategic timing rather than rushed reactions.

Topics

Sociocultural | Legal and regulatory

Similar viewpoints

Both emphasized that successful content strategy should prioritize meaningful engagement and authentic community connections over vanity metrics, focusing on quality interactions and creative approaches to make factual information accessible and engaging.

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax

Arguments

Focus on quality of engagement and comments rather than follower counts or view numbers when measuring content success

Successful content requires being creative, partnering with influencers, uplifting community voices, and making facts entertaining while remaining educational

Topics

Sociocultural | Human rights

Both humanitarian organizations recognized that they operate in highly polarized, conflict-affected environments where misinformation has direct, severe consequences including violence and displacement, and where they face well-resourced opposition.

Speakers

– Gisella Lomax
– Ghaleb Cabbabe

Arguments

Information risks directly cause real-world harm including violence, killings, and forced displacement, as seen with Rohingya refugees in Myanmar

Organizations face opponents with significant resources including state actors and misinformation farms operating in highly polarized environments

Topics

Human rights | Cybersecurity

Both emphasized the intensive, multi-layered verification and review processes required to ensure information accuracy, particularly in sensitive contexts, while making the final product appear accessible and simple to audiences.

Speakers

– Ghaleb Cabbabe
– Eeva Moore

Arguments

Information verification requires multiple levels of checking and cross-checking, especially when dealing with sensitive situations involving colleague safety

Expertise must be baked into the production process with multiple layers of review, making content appear simple while requiring heavy behind-the-scenes work

Topics

Legal and regulatory | Human rights

Unexpected consensus

Strategic patience in responding to misinformation rather than immediate reaction

Speakers

– Ghaleb Cabbabe
– Eeva Moore

Arguments

Balance speed with thoroughness by assessing risks before communicating, sometimes taking time to understand how misinformation evolves rather than rushing responses

Time constraints create fundamental challenges since disruption moves faster than constructive response, requiring creative solutions and pre-prepared resources

Explanation

Despite the fast-paced nature of social media and the pressure to respond quickly to misinformation, both speakers advocated for strategic patience and thorough preparation over immediate reactions. This is unexpected given the common assumption that social media requires instant responses.

Topics

Sociocultural | Legal and regulatory

The need for institutional trust alongside individual credibility

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax

Arguments

Healthcare professionals must maintain accountability and use real names to build credibility, avoiding commercial endorsements that could damage trust

Build partnerships with tech companies, academic institutions, and civil society organizations to leverage expertise and resources that humanitarian organizations lack

Explanation

While much discussion around social media focuses on individual influencers and personal branding, both speakers emphasized the continued importance of institutional credibility and partnerships, suggesting that individual and institutional trust must work together rather than compete.

Topics

Human rights | Sociocultural | Legal and regulatory

Overall assessment

Summary

The speakers demonstrated strong consensus on the transformative potential of digital platforms for positive information dissemination, the serious nature of misinformation threats, and the need for multi-stakeholder, expertise-driven approaches to content creation and verification.

Consensus level

High level of consensus with complementary perspectives rather than conflicting viewpoints. The agreement spans across different sectors (healthcare, humanitarian, technology, civil society) suggesting broad applicability of these principles. This consensus indicates that despite different organizational contexts, there are shared challenges and effective strategies that can be applied across sectors for information integrity.

Different viewpoints

Speed vs. Thoroughness in Response Strategy

Speakers

– Dr. Ahmed Ezzat
– Ghaleb Cabbabe
– Eeva Moore

Arguments

Misinformation creates an unfair defensive position where evidence-based voices must defend against accusations while being constrained by factual accuracy

Balance speed with thoroughness by assessing risks before communicating, sometimes taking time to understand how misinformation evolves rather than rushing responses

Time constraints create fundamental challenges since disruption moves faster than constructive response, requiring creative solutions and pre-prepared resources

Summary

Dr. Ahmed emphasizes the defensive disadvantage of being constrained by evidence when responding to misinformation, while Ghaleb advocates for strategic patience and taking time to assess situations. Eeva focuses on the need for speed through creative solutions and pre-preparation, representing different approaches to the time-versus-accuracy dilemma.

Topics

Sociocultural | Legal and regulatory

Unexpected differences

Platform Algorithm Engagement Strategy

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax

Arguments

Focus on quality of engagement and comments rather than follower counts or view numbers when measuring content success

Successful content requires being creative, partnering with influencers, uplifting community voices, and making facts entertaining while remaining educational

Explanation

While both work on the same platform (TikTok), Dr. Ahmed advocates for focusing on engagement quality over metrics, while Gisella emphasizes creative partnerships and entertainment value. This represents different philosophies about how to effectively use social media algorithms – organic engagement versus strategic content optimization.

Topics

Sociocultural

Overall assessment

Summary

The panel showed remarkable consensus on core challenges and goals, with disagreements primarily centered on tactical approaches rather than fundamental principles. Main areas of difference included response timing strategies and platform engagement methods.

Disagreement level

Low to moderate disagreement level. The speakers demonstrated strong alignment on identifying challenges (misinformation threats, resource constraints, need for expertise) and broad goals (information integrity, community protection, credible content creation). Disagreements were primarily tactical and complementary rather than contradictory, suggesting different but potentially compatible approaches to shared challenges. This level of agreement is significant as it indicates a mature understanding of the field where practitioners can focus on refining methods rather than debating fundamental approaches.

Partial agreements

Similar viewpoints

Both emphasized that successful content strategy should prioritize meaningful engagement and authentic community connections over vanity metrics, focusing on quality interactions and creative approaches to make factual information accessible and engaging.

Speakers

– Dr. Ahmed Ezzat
– Gisella Lomax

Arguments

Focus on quality of engagement and comments rather than follower counts or view numbers when measuring content success

Successful content requires being creative, partnering with influencers, uplifting community voices, and making facts entertaining while remaining educational

Topics

Sociocultural | Human rights

Both humanitarian organizations recognized that they operate in highly polarized, conflict-affected environments where misinformation has direct, severe consequences including violence and displacement, and where they face well-resourced opposition.

Speakers

– Gisella Lomax
– Ghaleb Cabbabe

Arguments

Information risks directly cause real-world harm including violence, killings, and forced displacement, as seen with Rohingya refugees in Myanmar

Organizations face opponents with significant resources including state actors and misinformation farms operating in highly polarized environments

Topics

Human rights | Cybersecurity

Both emphasized the intensive, multi-layered verification and review processes required to ensure information accuracy, particularly in sensitive contexts, while making the final product appear accessible and simple to audiences.

Speakers

– Ghaleb Cabbabe
– Eeva Moore

Arguments

Information verification requires multiple levels of checking and cross-checking, especially when dealing with sensitive situations involving colleague safety

Expertise must be baked into the production process with multiple layers of review, making content appear simple while requiring heavy behind-the-scenes work

Topics

Legal and regulatory | Human rights

Key takeaways

Digital platforms like TikTok offer unprecedented reach for credible information – medical professionals can reach millions within 24 hours at zero cost, making them powerful tools for public health and humanitarian communication

Trust and credibility are built over time through consistent, accountable content creation – healthcare professionals must use real names and avoid commercial endorsements that could damage credibility

Effective content creation requires extensive behind-the-scenes work – multiple layers of expertise, fact-checking, and community input must be ‘baked into’ the production process to make content appear simple

Misinformation creates asymmetric warfare where bad actors with significant resources (including state actors) can move faster than those trying to provide accurate information

Quality of engagement matters more than follower counts – focus should be on meaningful comments and interactions rather than vanity metrics

Information risks cause real-world harm including violence, displacement, and erosion of trust in institutions, as demonstrated by the Myanmar Rohingya crisis

Partnerships are essential – humanitarian organizations, tech companies, academic institutions, and civil society must collaborate to leverage complementary expertise and resources

Time is the fundamental challenge – balancing the need for prompt response with thorough fact-checking and risk assessment

Resolutions and action items

UNHCR invited collaboration from academics, tech companies, and NGOs, specifically requesting partnerships to fill research gaps and develop strategies

Gisella Lomax promoted a Wednesday 2 PM event showcasing their South Africa pre-bunking project testing inoculation theory

Recommendation to check UNHCR’s information integrity toolkit by Googling ‘UNHCR information integrity’

Suggestion for audience members to examine comment sections on Red Cross digital platforms to understand the complexity of misinformation challenges

Unresolved issues

How to adequately resource content moderation in less common languages, especially in volatile contexts where platforms don’t have business incentives

The ongoing challenge of weakening trust and safety capacities from some digital platforms and tech companies

How to effectively combat well-resourced misinformation operations including state actors and misinformation farms

The fundamental time asymmetry between those creating misinformation and those trying to counter it with factual information

How to maintain institutional trust while individual voices gain prominence in information sharing

Balancing the need for speed in digital communication with thorough fact-checking and risk assessment processes

Suggested compromises

Balance speed with thoroughness by preparing reactive strategies and scenario planning in advance, allowing for quick but informed responses

Focus on proactive content creation to fill information voids rather than only reactive responses to misinformation

Leverage partnerships to share the resource burden – organizations should collaborate rather than trying to build all capabilities internally

Accept that sometimes it’s better to wait and assess how misinformation evolves rather than rushing to respond immediately

Combine individual creator authenticity with institutional expertise – use real names and personal accountability while maintaining organizational standards

Actually, members of the public, regardless of what their understanding is, bearing in mind in the UK the reading age is about six or seven, the maths age is about three to four, they are phenomenally intuitive. They can sniff out right from wrong. You just need to be able to explain that information in an easy and simple way.

Speaker

Dr. Ahmed Ezzat

Reason

This comment challenges the condescending assumption that the public lacks intelligence or intuition about information quality. It reframes the problem from ‘people are gullible’ to ‘experts need to communicate better,’ which is a fundamental shift in perspective about information integrity challenges.

Impact

This comment shifted the discussion from focusing on platform mechanics to emphasizing the importance of respectful, accessible communication. It influenced the later emphasis on building trust through authentic engagement rather than top-down messaging.

Information risks such as hate speech and misinformation are directly and indirectly causing real world harm. And I mean violence, killings, persecution. It can even be a factor in forced displacement, in causing refugees… as we saw in Myanmar back in 2016, 2017, when hate speech… had a decisive role in the displacement, I think, of 700,000 Rohingya refugees into Bangladesh who are still there today.

Speaker

Gisella Lomax

Reason

This comment elevated the stakes of the discussion by connecting online misinformation to concrete, devastating real-world consequences. It moved beyond abstract concerns about ‘information integrity’ to demonstrate how digital platform content can literally displace populations and cause humanitarian crises.

Impact

This dramatically shifted the tone and urgency of the conversation. It transformed the discussion from a technical/marketing challenge to a humanitarian imperative, influencing subsequent speakers to emphasize the life-or-death importance of their work and the need for stronger partnerships and resources.

You could boil the challenge down perhaps to one word, and that’s time. If you’re in the business of trying to disrupt or break something, you get to move at a much faster pace than if you’re on the other side of that equation.

Speaker

Eeva Moore

Reason

This insight crystallizes a fundamental asymmetry in information warfare – that destructive actors have inherent speed advantages over those trying to build trust and provide accurate information. It’s a profound observation about the structural challenges facing legitimate information providers.

Impact

This comment provided a unifying framework for understanding the challenges all panelists had described. It led to the moderator’s closing reflection on the ‘tension between balancing that limited resource of time’ and influenced the final discussion about strategic timing in responses.

We have seen, to our dismay, a weakening of these capacities and perhaps less resourcing from some companies. And I would extend that, for example, to content moderation in less common languages… Are you adequately providing content moderation in less common languages in these very volatile contexts where you don’t actually have a business argument?

Speaker

Gisella Lomax

Reason

This comment exposed a critical gap in platform responsibility – the tendency to under-resource content moderation in markets that aren’t profitable, even when those markets may be experiencing the most severe consequences of misinformation. It challenges the business model underlying content moderation.

Impact

This introduced a more critical perspective on platform responsibility that hadn’t been present earlier in the discussion, adding complexity to what had been a more collaborative tone between content creators and platforms.

It should be labor-intensive. I mean, dealing with these types of issues when we’re creating them, when we want to be accurate, should look like a light lift, but actually, in fact, be a pretty heavy one… you have to be consuming a lot of information. Sadly, that includes coming across the disinformation, if you’re going to understand it and to understand how it’s navigating the space.

Speaker

Eeva Moore

Reason

This comment reveals the hidden complexity behind seemingly simple social media content and acknowledges the psychological toll of constantly engaging with misinformation. It challenges the expectation that good content should be quick and easy to produce.

Impact

This deepened the discussion about resource allocation and the true cost of maintaining information integrity, supporting other panelists’ calls for more institutional support and partnership approaches.

Overall assessment

These key comments fundamentally elevated and reframed the discussion from a tactical conversation about social media best practices to a strategic dialogue about asymmetric information warfare and humanitarian responsibility. Dr. Ahmed’s insight about public intuition shifted the focus from platform mechanics to communication respect and accessibility. Gisella’s Myanmar example transformed the stakes from abstract ‘information integrity’ to concrete life-and-death consequences, while her critique of platform resource allocation introduced necessary tension about corporate responsibility. Eva’s ‘time’ framework provided a unifying theory for understanding the structural disadvantages faced by legitimate information providers. Together, these comments created a progression from individual content creation strategies to systemic analysis of power imbalances, resource constraints, and humanitarian imperatives in the digital information ecosystem. The discussion evolved from ‘how to create good content’ to ‘how to address fundamental inequities in information warfare while serving vulnerable populations.’

How can digital platforms adequately provide content moderation in less common languages in volatile contexts where there’s no business argument?

Speaker

Gisella Lomax

Explanation

This addresses a critical gap in platform safety measures for vulnerable populations who communicate in languages that aren’t commercially viable for platforms but still need protection from harmful content

What are the research gaps in addressing information integrity challenges in humanitarian contexts?

Speaker

Gisella Lomax

Explanation

Academic research is needed to inform strategies and policies for combating misinformation and hate speech that directly harms refugee and displaced populations

How can the effectiveness of ‘pre-bunking’ or inoculation theory be measured in building community resilience against hate speech and misinformation?

Speaker

Gisella Lomax

Explanation

UNHCR is testing this approach in South Africa but more research is needed to understand its effectiveness and scalability across different contexts

What metrics should be prioritized when evaluating content quality beyond traditional engagement metrics like followers and likes?

Speaker

Dr. Ahmed Ezzat

Explanation

Understanding how to measure meaningful engagement and content impact is crucial for organizations trying to combat misinformation effectively

How can institutions maintain public trust in an era where people increasingly distrust official sources?

Speaker

Gisella Lomax

Explanation

This addresses the broader challenge of institutional credibility when individual voices are often more trusted than official organizations

What are the resource implications and funding challenges for organizations trying to combat misinformation?

Speaker

Bia Barbosa (audience member)

Explanation

The question was cut off in the transcript but relates to the resource constraints mentioned by multiple panelists in fighting well-funded misinformation campaigns

How can partnerships between humanitarian organizations, tech companies, and civil society be structured to effectively address information integrity challenges?

Speaker

Gisella Lomax

Explanation

Given the multifunctional nature of this work and limited resources, understanding effective partnership models is crucial for scaling solutions

Summary

This discussion centered on the Freedom Online Coalition’s updated Joint Statement on Artificial Intelligence and Human Rights for 2025, presented during a session on shaping global AI governance through multi-stakeholder action. The panel featured representatives from Estonia, the Netherlands, Germany, Ghana, and Microsoft, highlighting the collaborative effort to address AI’s impact on human rights.

Ambassador Ernst Noorman from the Netherlands emphasized that human rights and security are interconnected, noting that when rights are eroded, societies become more unstable rather than safer. He stressed that AI risks are no longer theoretical, citing examples of AI being used to suppress dissent, distort public discourse, and facilitate gender-based violence. The Netherlands learned this lesson through their painful experience with biased automated welfare systems that deepened injustice for citizens.

The panelists identified surveillance, disinformation, and suppression of democratic participation as the most urgent human rights risks posed by AI, particularly when embedded in government structures without transparency or accountability. Dr. Erika Moret from Microsoft outlined the private sector’s responsibilities, emphasizing adherence to UN guiding principles on business and human rights, embedding human rights considerations from design through deployment, and ensuring fairness and inclusivity in AI systems.

Several binding frameworks were discussed as important next steps, including the EU AI Act, the Council of Europe’s AI Framework Convention, and the Global Digital Compact. The panelists emphasized the importance of multi-stakeholder collaboration, procurement policies that prioritize human rights-compliant AI systems, and the need for transparency and accountability in high-impact AI applications. The statement, endorsed by 21 countries at the time of the session, remains open for additional endorsements and represents a principled vision for human-centric AI governance grounded in international human rights law.

Keypoints

## Major Discussion Points:

– **Launch of the Freedom Online Coalition’s Joint Statement on AI and Human Rights 2025**: The primary focus was presenting an updated statement led by the Netherlands and Germany that establishes human rights as the foundation for AI governance, with 21 countries endorsing it and more expected to follow.

– **Urgent Human Rights Risks from AI**: Key concerns identified include arbitrary surveillance and monitoring by governments, use of AI for disinformation and suppression of democratic participation, bias and discrimination against marginalized groups (especially women and girls), and concentration of power in both state and private sector hands without adequate oversight.

– **Multi-stakeholder Responsibilities and Collaboration**: Discussion emphasized that addressing AI governance requires coordinated action from governments (through regulation and procurement policies), private sector (through human rights due diligence and ethical AI principles), and civil society, with Microsoft presenting their comprehensive approach as an example.

– **Binding International Frameworks and Next Steps**: Panelists highlighted several key governance mechanisms including the EU AI Act, Council of Europe’s AI Framework Convention, UN Global Digital Compact, and various UN processes, emphasizing the need for broader adoption and implementation of these frameworks.

– **Practical Implementation Challenges**: The discussion addressed real-world complexities of protecting human rights in repressive regimes, balancing innovation with rights protection, and the need for transparency while protecting privacy, with examples from the Netherlands’ own failures in automated welfare systems.

## Overall Purpose:

The discussion aimed to launch and promote the Freedom Online Coalition’s updated joint statement on AI and human rights, while building consensus among governments, private sector, and civil society on the urgent need for human rights-centered AI governance and identifying concrete next steps for implementation.

## Overall Tone:

The tone was consistently collaborative and constructive throughout, with speakers demonstrating shared commitment to human rights principles. There was a sense of urgency about AI risks but also optimism about multilateral cooperation. The discussion maintained a diplomatic yet determined quality, with participants acknowledging challenges while emphasizing collective action and practical solutions. The tone became slightly more interactive and engaging during the Q&A portion, with audience questions adding practical perspectives from different regions and sectors.

Speakers

– **Zach Lampell** – Senior Legal Advisor and Coordinator for Digital Rights at the International Center for Not-for-Profit Law; Co-chair of the Freedom Online Coalition’s Task Force on AI and Human Rights

– **Rasmus Lumi** – Director General for International Organizations and Human Rights with the Government of Estonia; Chair of the Freedom Online Coalition in 2025

– **Ernst Noorman** – Cyber Ambassador for the Netherlands

– **Maria Adebahr** – Cyber Ambassador of Germany; Co-chair of the Task Force on AI and Human Rights

– **Devine Salese Agbeti** – Director General of the Cyber Security Authority of Ghana

– **Erika Moret** – Director with UN and international organizations at Microsoft

– **Audience** – Multiple audience members who asked questions during the Q&A session

**Additional speakers:**

– **Svetlana Zenz** – Works on Asia region, focusing on bridging civil society and tech/telecoms in Asia

– **Carlos Vera** – From IGF Ecuador

# Freedom Online Coalition’s Joint Statement on AI and Human Rights 2025: A Multi-Stakeholder Discussion on Global AI Governance

## Executive Summary

This session focused on the launch of the Freedom Online Coalition’s updated Joint Statement on Artificial Intelligence and Human Rights for 2025, an update to the coalition’s original 2020 statement. The panel brought together representatives from Estonia, the Netherlands, Germany, Ghana, and Microsoft to discuss the statement and broader challenges in AI governance. At the time of the session, the statement had been endorsed by 21 countries, with expectations for additional endorsements from both FOC and non-FOC members.

The discussion centered on how AI systems pose risks to fundamental human rights including freedom of expression, right to privacy, freedom of association, and freedom of assembly. Speakers addressed the need for human rights to be embedded in AI development from the design phase, the importance of multi-stakeholder collaboration, and the role of international frameworks in governing AI systems.

## Key Participants and Their Contributions

**Rasmus Lumi**, Director General for International Organizations and Human Rights with the Government of Estonia and Chair of the Freedom Online Coalition in 2025, opened the session with remarks about AI-generated notes, noting humorously that the AI “did say all the right things” regarding human rights, which led him to wonder about AI’s understanding of these concepts.

**Ernst Noorman**, Cyber Ambassador for the Netherlands and co-chair of the FOC Task Force on AI and Human Rights, shared his country’s experience with automated welfare systems, stating: “In the Netherlands, we have learned this the hard way. The use of strongly biased automated systems in welfare administration, designed to combat fraud, has led to one of our most painful domestic human rights failures.” He emphasized that “Innovation without trust is short-lived. Respect for rights is not a constraint, it’s a condition for sustainable, inclusive progress.”

**Maria Adebahr**, Cyber Ambassador of Germany and co-chair of the FOC Task Force on AI and Human Rights, highlighted transnational repression as a key concern, noting: “AI unfortunately is also a tool for transnational repression… in terms of digital and AI, we are reaching here new levels, unfortunately.” She announced that Germany had doubled its funding for the Freedom Online Coalition to support this work.

**Devine Salese Agbeti**, Director General of the Cyber Security Authority of Ghana, provided perspective on bidirectional AI misuse, observing: “I have seen how citizens have used AI to manipulate online content, to lie against government, to even create cryptocurrency pages in the name of the president, etc. So it works both ways.” He identified surveillance, disinformation, and suppression of democratic participation as key concerns.

**Dr. Erika Moret**, Director with UN and international organisations at Microsoft, outlined private sector responsibilities, emphasizing that companies must adhere to UN guiding principles on business and human rights and embed human rights considerations throughout the AI development lifecycle.

**Zach Lampell**, co-chair of the FOC Task Force on AI and Human Rights, facilitated the discussion and emphasized that the statement remains open for additional endorsements, representing a principled vision for human-centric AI governance.

## Key Themes and Areas of Discussion

### Human Rights as Foundation for AI Governance

All speakers agreed that human rights should serve as the foundation for AI governance rather than being treated as secondary considerations. The discussion emphasized the need to put humans at the center of AI development and ensure compliance with international human rights law.

### Multi-Stakeholder Collaboration

Participants stressed that effective AI governance requires collaboration between governments, civil society, private sector, academia, and affected communities. Speakers noted that no single sector can address AI challenges alone.

### International Frameworks and Standards

The discussion covered several key governance mechanisms:

– **EU AI Act**: Referenced as creating predictability and protecting citizens, with potential for global influence

– **Council of Europe AI Framework Convention**: Presented as providing a globally accessible binding framework

– **UN Global Digital Compact**: Noted as representing the first time all UN member states agreed on an AI governance path

– **Hamburg Declaration on Responsible AI for the SDGs**: Mentioned as another relevant framework

– **UNESCO AI Ethics Recommendation**: Referenced in the context of global standards

### Urgent Human Rights Risks

Speakers identified several critical areas where AI poses immediate threats:

– Arbitrary surveillance and monitoring by governments

– Use of AI for disinformation campaigns and suppression of democratic participation

– Transnational repression using AI tools

– Bias and discrimination against marginalized groups

– Concentration of power without adequate oversight

### Private Sector Responsibilities

Dr. Moret outlined comprehensive responsibilities for companies, including:

– Adhering to UN guiding principles on business and human rights

– Embedding human rights considerations from design through deployment

– Conducting ongoing human rights impact assessments

– Ensuring transparency and accountability in AI systems

– Engaging in multi-stakeholder collaboration

## Practical Implementation Approaches

### Government Procurement

Ernst Noorman highlighted procurement as a practical tool, stating: “We have to use procurement as a tool also to force companies to deliver products which are respecting human rights, which have human rights as a core in their design of their products.”

### Coalition Building

Speakers discussed using smaller coalitions to achieve broader global adoption, with Noorman referencing the “oil spill effect” where regional frameworks like the Budapest Convention eventually gain wider acceptance.

### Diplomatic Engagement

The discussion emphasized combining formal diplomatic engagement with informal discussions to promote human rights principles in AI governance globally.

## Audience Questions and Broader Participation

The Q&A session included questions from both in-person and online participants. Carlos Vera from IGF Ecuador asked about how civil society organizations and non-FOC members can provide comments and support for the declaration. Svetlana Zenz raised questions about how risks can be diminished when tech companies work with oppressive governments and what practical actions civil society can take.

Online participants contributed questions about improving transparency in AI decision-making while protecting sensitive data, and identifying global frameworks with binding obligations on states for responsible AI governance.

## The FOC Joint Statement: Content and Next Steps

The statement addresses threats to fundamental freedoms including freedom of expression, right to privacy, freedom of association, and freedom of assembly. It provides actionable recommendations for governments, civil society, and private sector actors, addressing commercial interests, environmental impact, and threats to fundamental freedoms.

Several concrete next steps emerged from the discussion:

– The Task Force on AI and Human Rights committed to meeting to discuss creating space for civil society comments and support

– FOC members agreed to continue diplomatic engagement with non-member governments

– Continued work on promoting the statement’s principles through both formal and informal channels

## Implementation Challenges

The discussion acknowledged several ongoing challenges:

– Balancing transparency in AI decision-making with protection of sensitive data and privacy rights

– Addressing both government and citizen misuse of AI systems

– Ensuring meaningful participation of Global South countries and marginalized communities

– Protecting human rights in AI systems operating under repressive regimes

## Conclusion

The session demonstrated broad agreement among diverse stakeholders on the need for human rights-based AI governance. The FOC Joint Statement on AI and Human Rights 2025 provides a framework for coordinated international action, with concrete recommendations for implementation across sectors. The discussion emphasized practical approaches including government procurement policies, multi-stakeholder engagement, and diplomatic outreach to promote these principles globally.

The collaborative approach demonstrated in the session, combined with specific commitments to follow-up actions and broader engagement, positions the Freedom Online Coalition’s initiative as a significant contribution to global AI governance discussions. The success of this approach will depend on sustained commitment to implementation and continued collaboration across sectors and borders.

Zach Lampell: Welcome to the session, Shaping Global AI Governance Through Multi-Stakeholder Action, where we are pleased to present the Freedom Online Coalition’s Joint Statement on Artificial Intelligence and Human Rights 2025. My name is Zach Lampell. I’m Senior Legal Advisor and Coordinator for Digital Rights at the International Center for Not-for-Profit Law. I’m also pleased to be a co-chair of the Freedom Online Coalition’s Task Force on AI and Human Rights with the Government of the Netherlands and the Government of Germany. I want to welcome Mr. Rasmus Lumi, Director General for International Organizations and Human Rights with the Government of Estonia and the Chair of the Freedom Online Coalition in 2025.

Rasmus Lumi: Thank you very much. Good morning, everybody. It is a great honor for me to be here today to welcome you all to this session on the issue of how we feel or are deceived by the extremely smart artificial intelligence. When I read through the notes that it offered me, it did say all the right things, which is totally understandable. The question is, did it do it on purpose, maybe, maliciously, trying to deceive us into thinking that AI also believes in human rights? So we’ll have to take care of this. And this joint statement that we have developed under the leadership of the Netherlands is exactly one step in the way of doing this, putting humans in the center of AI development. So I would like to take this opportunity to very much thank the Netherlands for leading this discussion, this preparation in the Freedom Online Coalition, and I hope that the coalition and also elsewhere, this work will continue in order to make sure that humans and human rights will remain in the focus of all technological development. Thank you very much.

Zach Lampell: Thank you. I now turn the floor to Ambassador Ernst. Noorman, the Cyber Ambassador for the Netherlands.

Ernst Noorman: Thank you very much, Zach, and thank you, Rasmus, for your words. While leaders at this moment gather in The Hague to discuss defence and security, we are here to address a different but equally urgent task, protecting human rights in the age of AI. These are not separate. Human rights and security are, or should be, two sides of the same coin. When rights are eroded, when civic space shrinks, when surveillance escapes oversight, when information is manipulated, societies don’t become safer. They become more unstable, more fragile. Since the original FOC statement on AI and human rights in 2020, a lot has happened. I only have to mention the introduction of Chad Gipetit, just referred to by Rasmus, in November 2022, and how different AI tools are evolving every single day. It is shaping governance, policy, and daily life. Its benefits are real, but so are the risks. And those risks are no longer theoretical. We now see AI used to represent dissent, distort public discourse, and facilitate gender-based violence. In some countries, these practices are becoming embedded in state systems, with few checks and less or no transparency. At the same time, only a handful of private actors shape what we see. They influence democratic debate, and dominate key markets, without meaningful oversight. This double concentration of power threatens both public trust and democratic resilience. That’s why the Netherlands, together with Germany, and the International Centre for Not-for-Profit Law, has led the update of the Freedom on the Line Coalition’s joint statement on artificial intelligence and human rights. I’m grateful to all of you, governments, civil society, private sector, experts, for the thoughtful contributions that shaped it. This updated statement of our joint response to the present reality of AI sets out a principled and practiced vision. Vision for a human-centric AI, governed with care, grounded in human rights, and shaped through inclusive multi-stakeholder processes. It recognizes that risks arise across the AI lifecycle, not only through misuse, but from design to deployment. And the statement calls for clear obligations for both states and private sector, strong safeguards for those most at risk, especially women and girls. It calls for transparency and accountability in high-impact systems, and for cultural and linguistic inclusion, and attention should be given to the environmental and geopolitical dimensions of AI. Some claim that raising these issues could hinder innovation. We disagree. Innovation without trust is short-lived. Respect for rights is not a constraint, it’s a condition for sustainable, inclusive progress. In the Netherlands, we have learned this the hard way. The use of strongly biased automated systems in welfare administration, designed to combat fraud, has led to one of our most painful domestic human rights failures. It showed how algorithms, if not designed and deployed correctly, can deepen injustice. And it takes already years trying to correct the personal harm it caused. In response, we have strengthened our approach to avoid similar accidents, by applying human rights impact assessments, by applying readiness assessment methodology for AI human rights with the UNESCO, and by launching a national algorithm registry, with now more than a thousand algorithms being registered. But no country can solve this alone. All AI transcends borders. So must be our response. As of today, right now, 21 countries have endorsed this joint statement, and we expect more in the days ahead. The text will be published after this session, and remain open for further endorsements, including from non-FOC countries. Let us not stand idly by while others define the rules. Let us lead, clearly, collectively, and with conviction. Human rights must not be an afterthought in AI governance. They must be the foundation. Thank you very much.

Zach Lampell: Thank you, Ambassador Noorman. I think you’re absolutely right. Human rights needs to be the foundation of AI governance, and that was precisely what TFAIR, the Task Force on AI and Human Rights within the FOC, wanted to do with this joint statement. We wanted to build on previous statements and also make sure that there is a strong foundation for governance principles now with actionable, clear recommendations for governments, civil society, and the private sector. We’re really, really pleased with the joint statement and again want to thank the Netherlands and Germany for their co-leadership with ICNL and all of the TFAIR members and Freedom Online Coalition members for their support and input to the statement. We have an amazing, excellent panel today. I want to briefly introduce them and then we’ll get into questions. So joining us virtually is Maria Adebar, Cyber Ambassador of Germany, again co-chair of the Task Force on AI and Human Rights. To my right, Mr. Devine Salese-Agbeti, Director General of the Cyber Security Authority of Ghana. And next to him is Dr. Erika Moray, Director with UN and international organizations at Microsoft. So the first question, and this is directed at Devine. What is, in your view, the most urgent human rights risks posed by AI that this statement addresses?

Devine Salese Agbeti: Thank you very much. Firstly, I would like to thank the government of the Netherlands and also the FOC support unit for extending an invitation. Thank you for the invitation to Ghana for, or to participate in such an important conversation. In my view, I think the most urgent human risk, or the most urgent human right risk posed by AI is the arbitrary use of AI for monitoring or surveillance and also the use of AI for disinformation purposes and also for the suppression of democratic participation, particularly when such is embedded in a government structure and within law enforcement systems without any transparency and accountability. Those two are very important. When we look at these concerns, I think the broader fear is that when this is unchecked and also when they are governed by commercial interests, then what happens is that they erode fundamental freedoms or fundamental human rights, such as the freedom of speech and also privacy. And I think these are the key human right concerns when it comes to artificial intelligence.

Zach Lampell: Great, thank you. And several of those are specifically addressed within the statement, including commercial interests, the environmental impact of artificial intelligence, as well as the direct threats to fundamental human rights, such as the freedom of expression, the right to privacy, freedom of association, freedom of assembly, and others. Ambassador Adebar, thank you so much for joining us. What convinced you and the government of Germany to support this statement and what do you hope it will achieve?

Maria Adebahr: Hey, hello everybody over there. I hope you can hear me very well. Thanks for having me today. It’s a wonderful occasion to present and introduce our joint statement on artificial intelligence and human rights together. Thank you for the opening remarks and thank you Ernst together as a T-Fair co-chair. I really would like to thank you all for coming and joining us in this open forum session. So having you all here, I think it’s a really important sign of commitment for human rights and the broad multi-stakeholder participation these days. And that is even to become more important and let me explain why in doing this. Sometimes it helps to really go back to ask ourselves why do we do this and what led the government of Germany to support the statement. And this is the very essence that AI stands out as one of the most, if not the most transformative innovation and challenge and technological thing that we have to confront. And it will, it already does, change the way we live, work, express and inform ourselves. And it will change the way how we form our opinion and exercise our democratic rights. And as already Ernst said, in times of global uncertainty it offers a lot of promise but also a lot of risks to people on the planet. And that is why we as countries joining T-Fair and other forum have to answer the question in what kind of a digital future we want to live in. And I have to say, that’s really the essence of a human-centered world with a non-negotiable respect for human rights is what we have to strive for. And this is the essence that the statement gives us. And let me quote, it is a world firmly rooted in and in compliance with international law, including international human rights law, not shaped by authoritarian interests or solely by commercial priorities. And only with a wise international governance, we can harness the promise that the technologies as AI give to us and hold the horn at bay. And therefore, it is essential to us to support the statement and to support its principles, because we must stand with a strong focus on human rights and a commitment to a human-centric, safe, secure and trustworthy approach to technology. And as we all know, this is not a given yet, not anywhere in the world. So we hope to convince countries, civil societies and stakeholders to strive in every part of the world for our approach. This is crucial and this is very much the essence of what we have to do. And I’m also, as Ernst just said, very happy to have now 21 states on board. This is, I think, the majority and this gives our position even more weight. And also, let me mention the right of women and girls and all their diversity belong to groups posed to more vulnerability by AI. This is a strong commitment that we really clearly harnessed and wanted to see in there. So let me close by, yeah, I’m looking forward to questions and answers, obviously, but let me close by saying that I’m also very happy to announce that Germany is able to double its funding for the Freedom Online Coalition with the double amount compared to last year. We work through our budgets negotiations here in Germany, and so we were able to double the amount. And this is also something that makes me very happy and my colleagues. And hopefully we will make good use of our support for the Freedom Online Coalition. And with that heavy note, I get back over to you.

Zach Lampell: Thank you. Thank you, Ambassador, and thank you for doubling the funding for the Freedom Online Coalition. I can speak, I’m also a member of the Freedom Online Coalition’s advisory network, and I can say that we all believe that the FOC is a true driving force and leading vehicle to promote and protect fundamental freedoms and digital rights. And without the FOC, the governance structures that we have today, we believe, would be much weaker. So we look forward to continuing to work with you, Ambassador, as well as the government of the Netherlands, as TFAIR, as well as all of the 42 other member states of the Freedom Online Coalition, to continue our important work. And we welcome your leadership, and as civil society, we look forward to working with you to achieve these aims for everyone. Dr. Moret, if I could turn to you. You mentioned the commercial interest, and that is indeed something that is noted in the joint statement. What responsibility do private tech companies have to prevent AI from undermining human rights?

Erika Moret: Thank you very much. Well, thank you, first of all, to the government of the Netherlands, to the FOC, Excellencies, Ambassadors. It’s a real pleasure to be here today. I’m currently working at Microsoft, but in my past life I was an academic and working at the UN, including on various issues in relating to human rights and international humanitarian law, and it’s a real honor to be here. So as the tech industry representative on the panel, I’d like to address how private sector companies like Microsoft view responsibilities in ensuring AI respects and protects human rights. We are very lucky at Microsoft to have a series of teams and experts working on these areas across the company that fall under the newly created Trusted Technology Group, and this includes the Office of Responsible AI, the Technology for Fundamental Rights Group, and the Privacy, Safety, and Regulatory Affairs Group. So I’ll try and represent some of our collective work here in this area. In brief, we recognize that we must be proactive and diligent at every step in AI use, from design to deployment and beyond, to prevent AI from being used to violate human rights. The first is adherence to international standards. Microsoft and many peers explicitly commit to the UN guiding principles on business and human rights as a baseline for their conduct across global operations. For us and other companies, this involves a policy commitment to respect human rights and an ongoing human rights due diligence process that enables to identify and remedy related human rights harms. These principles make clear that while states must protect human rights, companies have an independent responsibility to respect fundamental rights. The next step is to embed human rights from the very beginning. We have a responsibility to integrate human rights considerations into the design and development of AI systems and really paying attention to areas like has already been highlighted in relation to women and girls and other particularly vulnerable groups. By performing human rights assessments and monitoring, this enables us to identify risks and address them. The establishment and enforcement of ethical AI principles is a third very important area. At Microsoft, we have a clearly defined responsible AI principles which encompass fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability which guide all of our AI development and our teams must follow our responsible AI standards which is a company-wide policy that translates these principles into specific requirements. Beyond human rights due diligence that I’ve already mentioned, we also work on protecting privacy and data security. Safeguarding users’ data is a non-negotiable responsibility for our company and also for our peers. AI often involves big data so companies must implement privacy by design, minimizing data collection, securing data storage and ensuring compliance with privacy laws. The sixth area I’d like to highlight here is the vital importance of fairness and inclusivity. Tech firms have the responsibility for ensuring that their AI does not perpetuate bias and discrimination and through working with partners at the FOC and then across civil society, we can put into measure active safeguards and ongoing work to tackle challenges in this area. Again, I’d like to highlight the important note in the FOC statement that AI harms are especially pronounced for marginalized groups. Groups. Ensuring transparency and explainability would be my seventh point of what we should be taking into consideration here, so that people understand how decisions are made and can I identify potential challenges, but then also mitigation approaches. The final area I’d like to emphasize here is the need for collaboration. We’re of course facing a fragile moment in terms of multilateralism and geopolitical tensions around the world and collaboration across borders and across sectors has never been more important. Engaging through multistakeholderism in AI governance, AI regulation developments is as much our responsibility as anyone else’s. I think the more the private sector can be working with civil society and with academia to improve its own work in these areas and to contribute as well through things like red teaming and other types of reporting. This is a really important next step, I believe, in our collective work. Thank you very much.

Zach Lampell: Thank you, Dr. Murray. I think the holistic approach from Microsoft is a fantastic model that I hope other companies will adopt or model their systems and internal standards after, and I really do appreciate the final point on collaboration and next steps. With that, Ambassador Norman, what are the most important next steps to ensure AI systems promote human rights and development while avoiding harm?

Ernst Noorman: Well, thank you. There are a number of issues we can do as governments, as international community, but first and foremost, I think the importance of good, strong regulation. We know, of course, there’s an international discussion on that. There are different views across the ocean about that, but at the same time, you create a level playing field. It’s predictability that’s important. As long as companies also know what the rules are, what the guardrails are, you protect the citizens in Europe and create trust, and then also in the products. So I think with the EU AI Act, that’s a good example, and we put rights, transparency, and accountability at the core. It’s risk-based, not blocking innovation, and I think that’s a crucial step, and you see that also a lot of countries outside the European Union look with great interest to the EU regulations and see how they can adapt it. At the same time, we have to continue to work at multilateral level. The discussion of how to implement oversight, to organize oversight, and to to ensure a safe implementation of AI. We have the Council of Europe’s AI Framework Convention, which is the next step, just like we had before on Cybercrime, the Budapest Convention. And I think it’s very important to really support the UN’s work on human rights, especially the Office of the Human Rights Commissioner, the Office of the High Commissioner of Human Rights. It’s important to keep on supporting them. They have important initiatives, which are in BTEC, which also promote human rights in the private sector. So these are important steps we have to continue. And as Eric from Microsoft already mentioned, the UN Guiding Principles on Business and Human Rights is an important tool also for the private sector to be used. And finally, as an important step too, we as governments play an important role with procurement. So we have to use procurement as a tool also to force companies to deliver products which are respecting human rights, which have human rights as a core in their design of their products, and ensure that they provide safe products to the governments which are used broadly in the society. And I think that’s a very concrete step we can use as governments. Thank you very much.

Zach Lampell: Thank you, Ambassador. I think procurement is a really great final point to mention. Ambassador Adebar, same question to you. What are the most important next steps that we can take to ensure AI systems promote human rights and development while avoiding harm?

Maria Adebahr: Thank you. Yeah, thank you for the question. Ernst just answered it, I think. So we fully can agree with that Ernst just said. It is moving forward on a lot of fronts and different fora. That starts with the EU-AI Act and its implementation and the promotion of its principles. We know certain aspects are being discussed, especially by industries and worldwide, and this is totally okay and this is rightfully so. But we can start and have a discussion on things like risk management and so on and so forth. But what the EU-AI Act really sparks is a discussion on how we want to manage and govern ourselves with AI and with a good and human-centric AI accessible for everybody. So this is one part. As EU member states, UNESCO recommendations of 2021 are another one, and we are looking forward also to the results of the Third Global Forum on the Ethics of AI, Health and Banking. at this very moment. And also we would invite all FOC members and interest actors to join the Hamburg Declaration on Responsible AI for the SDGs, recently stated. And in implementation of the Global Digital Compact we’re currently discussing modalities resolution and modalities for implementing working groups, a panel and a worldwide dialogue on AI. This is very much important to us. And one final point, AI unfortunately is also a tool for transnational repression. And transnational repression is a phenomenon, let me put it like that, that we as German governments really want to focus more on. Because it’s a fairly not a new phenomenon, it’s very, very old. But in terms of digital and AI, we are reaching here new levels, unfortunately. And so this is also a subject we want to discuss more and bring it in international forward and on the international human rights agenda. Thank you.

Zach Lampell: Thank you, Ambassador. And Director Agbedi, same question to you. What are the most important next steps to ensure AI systems promote human rights and development?

Devine Salese Agbeti: Thank you. Firstly, we have to align AI with international human rights standards. In that, for example, currently the Cyber Security Authority is working with the Data Protection Commission in Ghana to actually explore redress for this. Secondly, we can look at the UN guiding principles on business and human rights. The ICCPR can underpin national, regional and international efforts to redress global AI governance in that aspect. And we can also look at the multi-stakeholder participation, which is foundational to this. The ecosystem must include civil society to amplify marginal voices, technical communities to bring transparency into algorithms, and academia to provide evidence-based insights, a private sector to ensure responsible innovation, as well as youth and indigenous voices to reflect world diversity. I think these are the ways to do so.

Zach Lampell: Fantastic. Thank you. Thank you so much. Thank you very much for again very comprehensive ideas from both ambassadors and Director Getty. Now we would like to open up the floor to all of you. We have some time for questions and answers. We have a microphone over to my left, your right. We welcome questions on the joint statement and how best to promote human rights and artificial intelligence.

Audience: Hello. It works. All right. Sorry. It’s like to handle all these devices. Hello, everyone. Thank you so much for such a great panel. And I’m super happy that Freedom Online Coalition together with the private sector is coming up with some, not all, I mean, I would not call it solutions, but at least some recommendations. My name is Svetlana Zenz, and I’m working on Asia region. And, I mean, basically my work is actually to bring the bridge between civil society and tech and telecoms in Asia. And also they’ve been mentioned today, BTAC project, which is also a great project and was like one of the start of many other initiatives in the sector. I really hope that all of that kind of initiatives will get together. But my question will be actually to FOC members, representatives, and to the private sector. So, for a start, I mean, we know that Microsoft works, I mean, known as a company which works closely with many governments because you have products which provide the operability of those governments. And in some countries as well. especially with oppressive laws and oppressive regimes, it’s hard to make sure that the human rights of users are protected. So do you have any vision how to diminish those risks? Maybe there should be more civil society action on that side, like being more practical on that side. For the OFOC side, we’ve been working with FOC members for several years starting from Myanmar where we were one of the first who engaged FOC members in this statement on internet shutdowns. FOC, what can you do in the side of, I mean, of course statement is great, but like in practical way, how we can make sure that human rights are protected on the physical world? Thank you.

Zach Lampell: Thank you so much, I think we’ll take one more question and then we’ll answer, and then if there’s time and additional questions, we’ll go back. So please, sir.

Audience: Thank you, good day. My name is Carlos Vera from IGF Ecuador. I read the declaration in the FOC website. It would be nice if we can have some space to comment on the declaration outside the FOC members and even to sign our support for that declaration. Some government even doesn’t know that the FOC exists, so maybe we can also create some warning in the civil society space. Thank you very much for the great work. Thank you.

Zach Lampell: Thank you so much. I think that’s a great suggestion and we actually have a meeting with the Task Force on AI and Human Rights tomorrow, and I will raise this very point and see what we can do to have wider adoption, especially from civil society support. So back to the first question, it’s really a question on how we, whether FOC governments, the private sector and potentially civil society can protect human rights. regarding AI systems, especially in repressive regimes. I hope I got that question correct. Okay, thank you. So anybody who would like to start us off?

Ernst Noorman: Maybe just to kick off, but I’m sure that others also want to contribute. This is online and offline dilemma. It’s not a specific, the fact that there are tools, of course, also by big tech companies like Microsoft, of course, gives them a responsibility to look at the guardrails of their tools. But at the same time, platforms are being used for human rights violations and to threaten people, but it happens offline as well. But the point is, as an FOC, wherever I go as a cyber ambassador, and I’m sure that Maria is doing the same, we discuss the role of FOC. So also with governments who are not a logical member of the FOC, we always explain what the agenda is of the FOC. Why is the FOC there? What are we doing? And why is it important, the topic on the agenda? You mentioned internet shutdowns. I can assure you I’ve been discussing that with many countries who use this tool as politically. And it’s inside the rooms. You can have more open discussion than if you do it only online with statements, et cetera. So it’s also important to have it inside rooms, closed-door sessions. Now, why are you using these tools? Can’t you do it, can’t you avoid this? Because you harm also a lot of civil services. You harm the role of journalists, which are crucial. So it’s one of the topics, just as an example, which I often mention, but also bringing up then, why is the FOC there? And the FOC is already since 2011. So it’s known by many governments, but also for new colleagues or new people working in the governments. We always stress the important role of it and also explain why it’s there and why it’s important to respect human rights online as it is important to do it offline.

Zach Lampell: Thank you, Ambassador. Ambassador Adelbardi, do you want to come in as well?

Maria Adebahr: Thank you so much. I can only underline what Ernst just said. It’s important, and we do our work and spreading the word and have those discussions formally, but also informally in a four-eye setting, politely, diplomatically sometimes, and other times really straightforward and forcefully is the way to go. And the FOC is a very, very important forum for doing this and a reference point for everybody of us, I think. Thank you.

Zach Lampell: Thank you, Ambassador. And maybe Dr. Moray, if I, oh, sorry. Director Gbedi, please.

Devine Salese Agbeti: All right, thank you. I think that as much as we are advocating for FOC members to ensure the human rights online, especially when it comes to AI, I think FOC should. We’ve been working on the Cyber Security Authority, I have seen how citizens have used AI to manipulate online content, to lie against government, to even create cryptocurrency pages in the name of the president, etc. So it works both ways. I think FOC should be advocating for responsible use of citizens, and at the same time when this is being advocated, then FOC can engage with government also to ensure that the citizens actually have the right to use these systems and use it freely without the fear of arrest or the fear of intimidation. Thank you.

Zach Lampell: Thank you. Dr. Moray.

Erika Moret: Thank you. It’s quite hard to build on these excellent points already. So just to add a few extra points on what has already been said. I would say that from the private sector perspective and from Microsoft viewpoint, we take a principled and proactive approach to this particular question, including due diligence before entering high-risk markets, which is guided by the UN guiding principles. We limit or decline services like facial recognition where misuse is likely. We also resist over-broad government data requests and publish transparency reports where misuse is likely, and to hold ourselves accountable, we offer tools like AccountGuard to protect civil society journalists and human rights defenders from cyber threats. And we also advocate globally for responsible AI and digital use, including through a very important process like the Global Digital Compact and also, of course, in fora such as the IGF and WSIS and so on. Also just to say I’d really like to highlight here the really important developments that have been going on in terms of AI data-driven tools to protect against human rights abuses under authoritarian regimes, and many tech companies are working proactively with the human rights community for this. We personally are very actively engaged with the Office of the High Commissioner on Human Rights across numerous different projects in terms of monitoring human rights abuses and helping to detect risks, and also in areas like capacity building and AI training in order to properly harness the tools and also come up with new solutions where needs are identified.

Zach Lampell: Thank you. Great answers to what is a very tough question and a really never-ending battle to prevent abuses or misuse of AI. We have a couple questions from our colleagues joining us online. The first is, what are any global frameworks with binding obligations on states for the responsible use and governance of AI? And the second question is, how can transparency in AI decision-making be improved without exposing sensitive data, particularly to ensure that the right to privacy is protected under international human rights law is indeed protected? So would any of the panellists like to jump in on either of those two questions? We have about five minutes left. Please, Dr. Murray.

Erika Moret: Okay, well, maybe I’ll just kick it off talking about the importance of the GDC that I already mentioned, the Global Digital Compact, and I’m sure everybody in the room knows about it already. It was launched back for the future at the last UN General Assembly, the first time every UN member state in the world came together to agree a path on AI governance, and I think it’s incredibly important. There are two new bodies that are being developed right now, Dialogue and the Panel, and Microsoft has been engaged at every step of the process, sitting at the table, and we’ve been very grateful to have a voice there. And I think the more private sector, but also, of course, civil society, particularly those without the usual access to these types of processes is incredibly vital, but not just to have a seat at the table, but to actually have a voice at the table. So the more we can find inclusive, fair, transparent, participatory ways that those, particularly in the global majority, can have a meaningful way of engaging in these very, very important developments through this multi-stakeholder model is to be encouraged in my view. Thanks.

Zach Lampell: Thank you. Anybody else? Ambassador?

Ernst Noorman: I would first give the floor to Marie-Anne before I take away.

Zach Lampell: Apologies.

Maria Adebahr: Oh, that’s kind, Ant. I would use this opportunity to, again, put your attention to the Council of Europe Framework Convention on AI and Human Rights, because really this is something that was hardly negotiated. It’s global, and we would strive for more member states to join. And it’s open for all. It’s a globally open convention. You don’t have to be a member of the Council of Europe. So please have a look or tell your state representatives, respectively, to have a look and join. You can always approach, I think, any of EU member states for more information. The second internationally binding or hopefully really to be implemented thing is the Global Digital Compact already mentioned. And it is important, I think, because our headcount, in our headcount, we came to the conclusion that this is really the only truly global forum to discuss AI. And if we wouldn’t do it there, then about more than 100 states worldwide would not be present at any important table to discuss AI governance, because those states are probably not a member of the G7, G20, UNESCO, OSCE, or not able, in terms of resources, to join those discussions in depth. So this makes the Global Digital Compact even more important. The EU-AI Act has, by its nature, aspects of AI governance and principles inherent. So this would be the third framework I would like to mention here. Thank you.

Ernst Noorman: And now I can just add a few points on what Maria said. What the EU-AI Act can have is what we call the Brussels effect. Maria mentioned already the Council of Europe’s AI Framework Convention. If we would strive for a binding framework within the whole UN, it’s going to be very difficult. But if you see these more smaller coalitions, I mentioned the Budapest Convention on Cybercrime, what we have seen, just during the negotiations on the UN Cybercrime Treaty, that more and more members from other regions decided to join the Budapest Convention. From the Pacific, from Africa, from other regions, they decided, well, we want to be part of the Budapest Convention, which is very effective, very concrete cooperation on this topic. So I think that’s also a good example of how we can work in smaller coalitions with the oil spill effect to conquer the world and with good, strong legislation as well for other countries.

Devine Salese Agbeti: Thank you. Excellent points made by everyone here so far on this. And I would just like to add the Palma process also, which promotes the responsible use of these emerging technologies, which includes the artificial intelligence and it requires members. So we encourage other states to sign up as well to it and also member states to implement the convention so that we can all encourage responsible use of these technologies. Thank you.

Zach Lampell: Thank you. There are some very important, very significant binding governance mechanisms, like the Convention on AI from the Council of Europe, and that really can mimic the Budapest Convention, which has become the leading authority for combating and preventing cybercrimes. So this is a bit of a call to action from civil society. Let us, the FOC member states and FOC advisory network, help you inform your governments on these processes, and let us be able to help you advocate for them to adopt, sign and enact them. So it’s been a fantastic panel so far. Ambassador Noorman, please, some closing remarks from you.

Ernst Noorman: Thank you very much, Zach, for moderating also this panel. First of all, the statement is online right now, so go to the website Freedom Online Coalition and just copy the statement, put it on the social media, spread the word. So that’s already important, and I would really like to thank everyone who had been involved in drafting the statement, both from the members of the Freedom Online Coalition as the advisory network, who played an extremely important and meaningful role in strengthening the statement. We had the in-person meetings, one of the first ones in the Rights Conference in February this year, and we had a number of online coalitions, so a lot of work has been put in drafting and strengthening the statement. So I would really like to thank already the countries who have decided to sign on on the statement, and I’m confident that many more will follow in the days and weeks to come. And finally, I really would like to thank, on behalf, I think, of Maria and her team, and Zach and your team, and my team from the Netherlands, to thank all of you, first of all, to be present, and those who had been involved in drafting the statement for your dedication, your work, and your shared purpose on this important topic. Thank you very much. Thank you. Thank you, everyone.

Agreement points

Human rights must be foundational to AI governance

Speakers

– Rasmus Lumi
– Ernst Noorman
– Maria Adebahr
– Devine Salese Agbeti

Arguments

Need for human-centric AI development that puts humans at the center of technological advancement

Human rights must be the foundation of AI governance, not an afterthought

Germany supports human-centered world with non-negotiable respect for human rights over authoritarian or commercial interests

AI systems must align with international human rights standards including ICCPR and UN guiding principles

Summary

All government representatives agree that human rights should be the central organizing principle for AI governance, not treated as secondary considerations. They emphasize putting humans at the center of AI development and ensuring compliance with international human rights law.

Topics

Human rights | Legal and regulatory

Multi-stakeholder approach is essential for AI governance

Speakers

– Ernst Noorman
– Devine Salese Agbeti
– Erika Moret

Arguments

AI governance requires principled vision grounded in human rights and shaped through inclusive multi-stakeholder processes

Multi-stakeholder participation must include civil society, technical communities, academia, private sector and marginalized voices

Collaboration across sectors through multistakeholder engagement is essential responsibility

Summary

Speakers agree that effective AI governance requires inclusive participation from governments, civil society, private sector, academia, and marginalized communities. They emphasize that no single sector can address AI challenges alone.

Topics

Human rights | Sociocultural

International frameworks and standards are crucial for AI governance

Speakers

– Ernst Noorman
– Maria Adebahr
– Devine Salese Agbeti
– Erika Moret

Arguments

Strong regulation like EU AI Act creates level playing field and predictability while protecting citizens

Need to promote EU AI Act principles and UNESCO recommendations globally

AI systems must align with international human rights standards including ICCPR and UN guiding principles

Global Digital Compact represents first time all UN member states agreed on AI governance path

Summary

All speakers support the development and implementation of international frameworks for AI governance, including the EU AI Act, Council of Europe Convention, UN frameworks, and other binding international agreements.

Topics

Legal and regulatory | Human rights

AI poses urgent risks to democratic processes and human rights

Speakers

– Ernst Noorman
– Maria Adebahr
– Devine Salese Agbeti

Arguments

AI is used to repress dissent, distort public discourse, and facilitate gender-based violence

AI can be a tool for transnational repression reaching new levels through digital means

Most urgent risks are arbitrary AI surveillance and use for disinformation to suppress democratic participation

Summary

Government representatives agree that AI systems are being actively misused to undermine democratic institutions, suppress dissent, and violate human rights, particularly through surveillance and disinformation campaigns.

Topics

Human rights | Cybersecurity

Similar viewpoints

Both European representatives see EU AI regulation as a model that should influence global AI governance standards and be promoted internationally.

Speakers

– Ernst Noorman
– Maria Adebahr

Arguments

EU AI Act can have Brussels effect influencing global standards

Need to promote EU AI Act principles and UNESCO recommendations globally

Topics

Legal and regulatory | Human rights

Both emphasize the importance of embedding human rights considerations into AI systems from the design phase and using economic incentives to promote responsible AI development.

Speakers

– Ernst Noorman
– Erika Moret

Arguments

Government procurement should be used as tool to force companies to deliver human rights-respecting products

Companies must be proactive at every step from design to deployment to prevent human rights violations

Topics

Human rights | Economic

Both promote specific international frameworks that are open to global participation and provide binding commitments for responsible AI governance.

Speakers

– Maria Adebahr
– Devine Salese Agbeti

Arguments

Council of Europe Framework Convention on AI provides globally open binding framework

Palma process promotes responsible use of emerging technologies including AI

Topics

Legal and regulatory | Human rights

Unexpected consensus

Private sector proactive responsibility for human rights

Speakers

– Ernst Noorman
– Erika Moret

Arguments

Government procurement should be used as tool to force companies to deliver human rights-respecting products

Companies must be proactive at every step from design to deployment to prevent human rights violations

Explanation

There is unexpected alignment between government and private sector perspectives on corporate responsibility, with both agreeing that companies should proactively embed human rights protections rather than waiting for government mandates.

Topics

Human rights | Economic

Citizens’ role in AI misuse

Speakers

– Devine Salese Agbeti
– Ernst Noorman

Arguments

Citizens misuse AI to manipulate content and create fraudulent materials, requiring responsible use advocacy

FOC members should engage governments through formal and informal diplomatic discussions

Explanation

There is consensus that AI governance challenges come not only from governments and corporations but also from individual citizens, requiring education and advocacy for responsible use by all stakeholders.

Topics

Human rights | Cybersecurity

Overall assessment

Summary

There is strong consensus among all speakers on core principles: human rights as foundation of AI governance, need for multi-stakeholder approaches, importance of international frameworks, and recognition of urgent AI-related threats to democracy and human rights.

Consensus level

High level of consensus with remarkable alignment across government, private sector, and civil society representatives. This suggests strong potential for coordinated international action on AI governance, with the FOC joint statement serving as a foundation for broader cooperation. The consensus spans both principles and practical implementation approaches, indicating mature understanding of AI governance challenges.

Different viewpoints

Unexpected differences

Scope of AI misuse concerns

Speakers

– Devine Salese Agbeti
– Other speakers

Arguments

Citizens misuse AI to manipulate content and create fraudulent materials, requiring responsible use advocacy

Explanation

While most speakers focused on government and corporate misuse of AI, Agbeti uniquely highlighted citizen misuse as a significant concern, arguing that FOC should advocate for responsible use by individuals rather than only focusing on state and corporate actors. This represents an unexpected broadening of the discussion beyond the typical focus on institutional actors

Topics

Human rights | Cybersecurity

Overall assessment

Summary

The discussion showed remarkably high consensus among speakers on fundamental principles and goals for AI governance, with only minor differences in emphasis and approach rather than substantive disagreements

Disagreement level

Very low disagreement level. The speakers demonstrated strong alignment on core issues including the need for human rights-centered AI governance, multi-stakeholder approaches, and international cooperation. The few differences that emerged were primarily about tactical approaches (domestic vs. global focus, government vs. citizen responsibility) rather than fundamental disagreements about principles or goals. This high level of consensus likely reflects the collaborative nature of developing the FOC joint statement and suggests strong potential for coordinated action on AI governance among these stakeholders.

Partial agreements

Similar viewpoints

Both European representatives see EU AI regulation as a model that should influence global AI governance standards and be promoted internationally.

Speakers

– Ernst Noorman
– Maria Adebahr

Arguments

EU AI Act can have Brussels effect influencing global standards

Need to promote EU AI Act principles and UNESCO recommendations globally

Topics

Legal and regulatory | Human rights

Both emphasize the importance of embedding human rights considerations into AI systems from the design phase and using economic incentives to promote responsible AI development.

Speakers

– Ernst Noorman
– Erika Moret

Arguments

Government procurement should be used as tool to force companies to deliver human rights-respecting products

Companies must be proactive at every step from design to deployment to prevent human rights violations

Topics

Human rights | Economic

Both promote specific international frameworks that are open to global participation and provide binding commitments for responsible AI governance.

Speakers

– Maria Adebahr
– Devine Salese Agbeti

Arguments

Council of Europe Framework Convention on AI provides globally open binding framework

Palma process promotes responsible use of emerging technologies including AI

Topics

Legal and regulatory | Human rights

Key takeaways

Human rights must be the foundation of AI governance, not an afterthought, requiring a human-centric approach that puts people at the center of technological development

The most urgent AI human rights risks include arbitrary surveillance, disinformation campaigns, suppression of democratic participation, and the concentration of power in few private actors

Private sector companies have independent responsibility to respect human rights through proactive measures from design to deployment, including adherence to UN guiding principles on business and human rights

Multi-stakeholder collaboration involving governments, civil society, private sector, academia, and marginalized communities is essential for effective AI governance

The FOC Joint Statement on AI and Human Rights 2025 has been endorsed by 21 countries and provides actionable recommendations for all stakeholders

Strong regulatory frameworks like the EU AI Act, Council of Europe AI Framework Convention, and Global Digital Compact provide important binding and non-binding governance mechanisms

Government procurement can be used as a powerful tool to ensure AI systems respect human rights by requiring companies to deliver compliant products

Resolutions and action items

The FOC Joint Statement on AI and Human Rights 2025 is now published online and remains open for additional endorsements from both FOC and non-FOC countries

Germany announced doubling its funding for the Freedom Online Coalition to support AI and human rights work

Task Force on AI and Human Rights will meet to discuss creating space for civil society comments and support for the declaration beyond FOC members

FOC members committed to continue diplomatic engagement with non-member governments through formal and informal discussions to promote human rights in AI

Participants agreed to spread awareness of the statement through social media and other channels

Promotion of existing binding frameworks like the Council of Europe AI Framework Convention and Global Digital Compact implementation

Unresolved issues

How to effectively protect human rights in AI systems operating under repressive regimes while companies maintain government contracts

Balancing transparency in AI decision-making with protection of sensitive data and privacy rights

Addressing the dual challenge of preventing both government misuse of AI and citizen misuse of AI for fraudulent purposes

Ensuring meaningful participation of Global South countries and marginalized communities in AI governance processes

Creating truly global binding frameworks for AI governance given the difficulty of achieving consensus among all UN member states

Determining optimal mechanisms for civil society engagement and support for AI governance initiatives beyond government-led processes

Suggested compromises

Using smaller coalitions and regional frameworks (like Council of Europe Convention) to achieve ‘Brussels effect’ or ‘oil spill effect’ for broader global adoption rather than waiting for universal UN consensus

Leveraging government procurement policies as a practical middle-ground approach to enforce human rights standards without requiring new legislation

Balancing innovation promotion with rights protection through risk-based regulatory approaches that don’t block technological advancement

Combining formal diplomatic engagement with informal discussions to address AI human rights concerns with non-compliant governments

Using existing frameworks like UN guiding principles on business and human rights as baseline standards while developing more specific AI governance mechanisms

When I read through the notes that it offered me, it did say all the right things, which is totally understandable. The question is, did it do it on purpose, maybe, maliciously, trying to deceive us into thinking that AI also believes in human rights?

Speaker

Rasmus Lumi

Reason

This opening comment immediately established a critical and philosophical tone by questioning AI’s apparent alignment with human values. It introduced the concept of AI deception and whether AI systems might manipulate human perception of their intentions, which is a sophisticated concern beyond basic functionality issues.

Impact

This comment set the stage for deeper philosophical discussions throughout the session about AI’s relationship with human values and the need for human-centered governance. It moved the conversation beyond technical implementation to fundamental questions about AI’s nature and trustworthiness.

Innovation without trust is short-lived. Respect for rights is not a constraint, it’s a condition for sustainable, inclusive progress.

Speaker

Ernst Noorman

Reason

This reframes the common narrative that human rights protections hinder technological innovation. Instead, it positions human rights as essential infrastructure for sustainable technological development, challenging the false dichotomy between innovation and rights protection.

Impact

This comment shifted the discussion from defensive justifications of human rights to a proactive business case for rights-based AI development. It influenced subsequent speakers to discuss practical implementation rather than theoretical benefits.

In the Netherlands, we have learned this the hard way. The use of strongly biased automated systems in welfare administration, designed to combat fraud, has led to one of our most painful domestic human rights failures.

Speaker

Ernst Noorman

Reason

This vulnerable admission of failure from a leading democratic nation provided concrete evidence of how AI systems can cause real harm even with good intentions. It demonstrated intellectual honesty and showed that human rights violations through AI are not just theoretical concerns or problems of authoritarian regimes.

Impact

This personal example grounded the entire discussion in reality and gave credibility to the urgency of the human rights framework. It influenced other speakers to focus on practical safeguards and accountability mechanisms rather than abstract principles.

AI unfortunately is also a tool for transnational repression. And transnational repression is a phenomenon… that we as German governments really want to focus more on. Because it’s a fairly not a new phenomenon, it’s very, very old. But in terms of digital and AI, we are reaching here new levels, unfortunately.

Speaker

Maria Adebahr

Reason

This comment introduced a geopolitical dimension that expanded the scope beyond domestic AI governance to international security concerns. It highlighted how AI amplifies existing authoritarian tactics across borders, making it a global security issue requiring international coordination.

Impact

This broadened the conversation from individual rights protection to collective security, influencing the discussion toward international cooperation mechanisms and the need for coordinated responses to AI-enabled authoritarianism.

I have seen how citizens have used AI to manipulate online content, to lie against government, to even create cryptocurrency pages in the name of the president, etc. So it works both ways. I think FOC should be advocating for responsible use of citizens, and at the same time when this is being advocated, then FOC can engage with government also to ensure that the citizens actually have the right to use these systems and use it freely without the fear of arrest or the fear of intimidation.

Speaker

Devine Salese Agbeti

Reason

This comment introduced crucial nuance by acknowledging that AI misuse is not unidirectional – citizens can also misuse AI against governments. It challenged the implicit assumption that only governments and corporations pose AI-related threats, while maintaining the importance of protecting legitimate citizen rights.

Impact

This balanced perspective shifted the conversation toward more sophisticated governance approaches that address multiple threat vectors while preserving democratic freedoms. It influenced the discussion to consider comprehensive frameworks rather than one-sided protections.

We have to use procurement as a tool also to force companies to deliver products which are respecting human rights, which have human rights as a core in their design of their products, and ensure that they provide safe products to the governments which are used broadly in the society.

Speaker

Ernst Noorman

Reason

This identified government procurement as a powerful but underutilized lever for enforcing human rights standards in AI development. It provided a concrete, actionable mechanism that governments can implement immediately without waiting for comprehensive international agreements.

Impact

This practical suggestion energized the discussion around immediate actionable steps, moving from abstract principles to concrete implementation strategies that other speakers could build upon.

Overall assessment

These key comments fundamentally shaped the discussion by establishing it as a sophisticated, multi-dimensional conversation rather than a simple advocacy session. Lumi’s opening philosophical challenge set an intellectually rigorous tone, while Noorman’s admission of Dutch failures provided credibility and urgency. The comments collectively moved the discussion through several important transitions: from theoretical to practical (through concrete examples), from defensive to proactive (reframing rights as enabling innovation), from domestic to international (through transnational repression), and from one-sided to nuanced (acknowledging citizen misuse). These interventions prevented the session from becoming a simple endorsement of the joint statement and instead created a substantive dialogue about the complex realities of AI governance, ultimately strengthening the case for the human rights framework by acknowledging and addressing its challenges.

How can transparency in AI decision-making be improved without exposing sensitive data, particularly to ensure that the right to privacy is protected under international human rights law?

Speaker

Online participant

Explanation

This question addresses the critical balance between AI transparency requirements and privacy protection, which is fundamental to human rights compliance in AI systems

How can civil society organizations and non-FOC members provide comments and support for the FOC AI and Human Rights declaration?

Speaker

Carlos Vera from IGF Ecuador

Explanation

This highlights the need for broader participation and engagement mechanisms beyond FOC members, including creating awareness about FOC’s existence among governments and civil society

How can risks be diminished when tech companies like Microsoft work with oppressive governments, and what practical actions can civil society take?

Speaker

Svetlana Zenz

Explanation

This addresses the practical challenges of protecting human rights when technology companies operate in countries with repressive regimes and the role of civil society in mitigation

How can the FOC engage with governments to ensure citizens have the right to use AI systems freely while also advocating for responsible citizen use of AI?

Speaker

Devine Salese-Agbeti

Explanation

This explores the dual challenge of preventing both government overreach and citizen misuse of AI technologies, requiring balanced advocacy approaches

What are the global frameworks with binding obligations on states for the responsible use and governance of AI?

Speaker

Online participant

Explanation

This seeks to identify existing legally binding international mechanisms for AI governance, which is crucial for understanding the current regulatory landscape

How can the ‘Brussels effect’ of the EU AI Act be leveraged to influence global AI governance standards?

Speaker

Ernst Noorman (implied)

Explanation

This explores how regional regulations like the EU AI Act can create spillover effects to influence global standards, similar to the Budapest Convention’s expansion

How can transnational repression through AI tools be better addressed on the international human rights agenda?

Speaker

Maria Adebahr

Explanation

This identifies the need for focused attention on how AI is being used as a tool for transnational repression, which represents a new dimension of human rights violations

Summary

This panel discussion focused on privacy preservation within the Fediverse, also known as the open social web, which enables interoperability between different social media platforms through protocols like ActivityPub. The session was moderated by Mallory Knodel from the Social Web Foundation and featured panelists from various organizations including the Data Transfer Initiative, Meta, and academic researchers specializing in digital competition law.

The discussion explored fundamental challenges in maintaining user privacy when data flows between interconnected but independently operated platforms. A key technical issue highlighted was that when users delete posts on one platform, those deletions may not propagate across all federated services that have already received the content. This creates complications for compliance with privacy regulations like GDPR, which grants users rights to delete their personal data.

Panelists examined how existing privacy laws apply to decentralized systems, noting that current regulations were designed for centralized platforms and may need adaptation for federated environments. The European Union’s Digital Markets Act was discussed as potentially expanding interoperability requirements from messaging services to social media platforms in future reviews.

The conversation addressed governance challenges in distributed systems, emphasizing the need for shared standards and trust mechanisms across multiple platforms and instances. Speakers highlighted the importance of user education about how federated systems work, as many users may not understand that their posts can travel beyond their original platform.

The panel concluded that while the Fediverse offers promising opportunities for user agency and platform choice, realizing privacy-preserving interoperability requires ongoing collaboration between platforms, standards bodies, regulators, and civil society to develop appropriate technical and governance frameworks.

Keypoints

## Major Discussion Points:

– **Privacy challenges in interoperable social media systems**: The panel explored how federated platforms like the Fediverse can preserve user privacy while enabling cross-platform interaction, including technical issues like post deletion across multiple instances and the complexity of maintaining user control over data that flows between different services.

– **Regulatory frameworks and compliance in decentralized environments**: Discussion focused on how existing privacy laws like GDPR and emerging regulations like the Digital Markets Act apply to interoperable social media, with particular attention to the challenges of enforcing data subject rights (deletion, portability, access) across distributed networks.

– **Technical implementation and standardization of interoperability**: The conversation addressed the practical aspects of building interoperable systems, including the role of standards bodies like W3C and IETF, the importance of shared governance models, and the need for trust mechanisms between different platforms and instances.

– **User experience and education in federated systems**: Panelists discussed the challenge of making complex interoperable systems accessible to average users, including onboarding processes, setting appropriate defaults, and helping users understand what happens to their data when it crosses platform boundaries.

– **Governance and responsibility distribution across diverse stakeholders**: The panel examined how to manage a decentralized ecosystem involving thousands of instances and platforms, including the development of shared norms, trust registries, and coordination mechanisms that don’t centralize control but provide necessary oversight.

## Overall Purpose:

The discussion aimed to explore how privacy can be preserved and user rights protected in interoperable social media environments, particularly the Fediverse, while addressing the technical, legal, and governance challenges that arise when data and interactions flow across multiple platforms and instances.

## Overall Tone:

The discussion maintained a collaborative and constructive tone throughout, with panelists building on each other’s points rather than debating. The conversation was technical but accessible, with speakers acknowledging the complexity of the issues while expressing cautious optimism about solutions. The tone became more interactive and engaged during the Q&A portion, with audience questions adding practical perspectives and real-world concerns to the theoretical framework established by the panel.

Speakers

– **Mallory Knodel** – Executive Director and co-founder of the Social Web Foundation, focused on building a multipolar Fediverse with emphasis on human rights and community building

– **Chris Riley** – Academic/Professor, co-author of “Regulating Code” book, works on interoperability and digital policy issues, currently based in Melbourne, Australia

– **Delara Derakhshani** – Director of Policy and Partnerships at the Data Transfer Initiative (DTI), focused on data portability and user agency in data transfers

– **Ian Brown** – Technologist and academic, security and privacy expert, co-author of “Regulating Code,” works on digital competition law reform and interoperability, particularly with EU Digital Markets Act

– **Audience** – Various audience members who asked questions during the Q&A session

– **Melinda Claybaugh** – Director of AI and Privacy Policy at Meta, working on Threads product and Fediverse interoperability

**Additional speakers:**

– **Edmund Chung** – From .Asia organization

– **Dominique Zelmercier** – From W3C (World Wide Web Consortium)

– **Caspian** – Audience member who asked about 10-year vision

– **Winston Xu** – From One Power Foundation in Hong Kong

– **Gabriel** – Audience member who asked about trust between instances

– Various other unnamed audience members who asked questions

# Privacy Preservation in the Fediverse: Panel Discussion Report

## Introduction and Panel Setup

This panel discussion was moderated by Mallory Knodel from the Social Web Foundation as part of an Internet Governance Forum (IGF) session examining privacy challenges in federated social media systems. The session brought together experts to discuss how user privacy can be maintained when data flows between independently operated but interconnected platforms using protocols like ActivityPub.

The panelists included:

– Ian Brown, discussing regulatory frameworks and interoperability requirements

– Melinda Claybaugh from Meta, representing industry perspectives on federated systems (her full introduction was cut off in the recording)

– Dominique Zelmercier from the W3C, bringing standards body expertise

– Delara Derakhshani from the Data Transfer Initiative, focusing on user data portability

– Chris Riley, contributing insights on user experience and interoperability myths

The discussion emerged against the backdrop of growing interest in decentralised social media alternatives, particularly following user migrations from centralised platforms, with Chris Riley noting people “noisy quitting X” and subsequently “quiet quitting Mastodon.”

## Core Technical Challenges

### Data Deletion and Cross-Platform Propagation

Mallory Knodel highlighted a fundamental technical problem: when users delete posts on one platform, those deletion requests may not effectively propagate across all federated instances that have already received the content. This creates complications for privacy regulation compliance, particularly with GDPR requirements for data deletion.

Ian Brown provided a concrete example of this challenge, describing a scenario where a user sets their Mastodon posts to auto-delete after two months but uses a bridge service (Bridgyfed) to share content with BlueSky. The user’s privacy preferences from Mastodon do not automatically transfer to BlueSky, creating gaps in privacy protection across the federated network.

### Bridge Services and Privacy Preferences

The discussion revealed significant challenges with bridge services that connect different federated platforms. Ian Brown noted that Bridgyfed, which bridges content between platforms, doesn’t currently recognise or apply user privacy preferences across different systems. This technical limitation demonstrates the gap between user expectations and current implementation reality.

### Standards and Implementation Complexity

Dominique Zelmercier from the W3C raised concerns about the granularity of privacy controls in standardised systems. They observed that standardisation processes typically produce simple binary signals—”do or do not”—which may be inadequately coarse for the complexity of social media interactions where users want nuanced control over their digital presence.

## User Experience and Understanding

### User Comprehension Challenges

Melinda Claybaugh emphasised that many users may not understand what it means to post content on federated platforms or comprehend how their content might be shared across networks. She noted that users posting on Threads, for instance, may not realise their content could be shared with other services in the federation.

The discussion revealed that different user types—power users, technical users, and average users—require different levels of explanation and control options. Claybaugh stressed the importance of meeting people where they are and designing systems that accommodate varying levels of technical literacy.

### The “Myth of the Superuser”

Chris Riley made a significant intervention challenging what he called the “myth of the superuser.” He argued that the goal should be making interoperability invisible to users rather than requiring them to become technical experts. Riley emphasised that most users want good defaults rather than complex configuration options, creating tension between user education approaches and user experience simplicity.

## Regulatory Framework Discussion

### Existing Privacy Laws and Federation

Ian Brown discussed how existing privacy regulations like GDPR apply to decentralised systems, noting these laws provide important legal backstops against bad actors who ignore user data preferences. However, he acknowledged that current regulations were designed with centralised platforms in mind.

Brown referenced GDPR Article 22 (though noting he was speaking from memory and would need to verify) regarding automated decision-making, highlighting how existing legal frameworks need fresh implementation approaches for federated environments.

### Digital Markets Act Implications

Ian Brown provided insights into the European Union’s Digital Markets Act (DMA), which currently includes messaging interoperability requirements for designated gatekeepers. He presented slides showing the complexity of regulatory requirements, which he acknowledged were “too complicated” and “impossible to read,” illustrating the challenge of implementing nuanced regulatory frameworks.

Brown suggested that regulatory frameworks are evolving to require dominant firms to enable interoperability, representing a shift from traditional competition law approaches.

## Governance and Instance-Level Management

### Defederation as Governance Tool

Mallory Knodel highlighted defederation—where instances disconnect from others—as a mechanism for protecting users from bad actors whilst maintaining overall system openness. This represents a form of distributed governance that allows communities to maintain their safety standards without requiring centralised oversight.

### Government and Institutional Roles

Knodel made a specific policy recommendation, suggesting that “maybe governments should have instances where they’re the source of truth that’s verified.” This reflects broader discussions about how institutional actors might participate in federated systems whilst maintaining credibility and user trust.

Ian Brown supported this direction, arguing that governments should share information across multiple platforms rather than forcing citizens to use single platforms for accessing public services.

## Audience Questions and Key Responses

### Digital Identity and Democratic Institutions

An audience member raised questions about digital identity ownership and its relationship to democratic institutions. The panelists discussed how federated systems might support more democratic approaches to digital identity management, though specific solutions remained largely theoretical.

### Technical Standards Bodies Role

Questions about the role of standards organisations like W3C and IETF revealed their crucial function in developing interoperability standards and verification processes. The discussion highlighted both the importance of these bodies and the limitations of current standardisation approaches for complex social media interactions.

### Youth and Internet Interoperability

An audience question about youth involvement in internet interoperability prompted discussion about generational differences in approaching federated systems and the importance of including diverse perspectives in governance discussions.

### Cultural Context and Platform Migration

A significant audience question addressed how cultural context changes when content moves between platforms with different community norms. This highlighted that interoperability involves not just technical challenges but also social and cultural considerations that affect user safety and community dynamics.

### Trust and Security Between Instances

Questions about trust between instances and security concerns revealed ongoing challenges in federated systems. The panelists acknowledged that security vulnerabilities, particularly around direct messages and encrypted content, require continued attention as federation scales.

## Data Portability and User Migration

### Beyond Technical Transfer

Delara Derakhshani provided crucial insights into the complexity of platform migration, noting that it involves not just data transfer but joining new communities with different safety considerations. For marginalised groups particularly, the safety of the destination platform matters as much as the ability to bring content.

This reframed interoperability from a purely technical challenge to a human-centred one, highlighting that data portability is fundamentally about community safety and user agency.

### Trust Registries and Verification

Derakhshani emphasised the need for trust registries and verification processes to reduce duplication and streamline onboarding across platforms. These mechanisms could provide coordination without centralising control, addressing fundamental challenges of distributed governance.

## Future Vision and Ongoing Challenges

### Long-term Interoperability Goals

Mallory Knodel described a future where users never have to sign up for new services whilst still being able to follow people across platforms. Ian Brown envisioned social media becoming as configurable and flexible as the internet at the IP layer.

### Persistent Technical Issues

Several technical challenges remain unresolved, including:

– Propagation of deletion requests across federated networks

– Security vulnerabilities in direct messages and encrypted content

– Maintaining consistent user experiences across platforms with varying capabilities

– Handling privacy controls and auto-deletion features across different systems

### Scaling Governance

As federated systems potentially scale to thousands of instances, new coordination mechanisms will be needed to maintain trust and shared standards. The panelists acknowledged that current approaches may not be adequate for the scale of federation envisioned.

## Key Takeaways

The discussion revealed both the promise and complexity of building privacy-preserving federated social media systems. While technical solutions for basic interoperability exist, significant challenges remain in:

1. **Privacy Preservation**: Ensuring user privacy preferences and deletion requests propagate effectively across federated networks

2. **User Understanding**: Helping users comprehend the implications of posting in federated environments without overwhelming them with technical complexity

3. **Governance at Scale**: Developing coordination mechanisms that maintain standards and user protections across distributed systems

4. **Regulatory Adaptation**: Implementing existing privacy laws in federated contexts while potentially developing new regulatory approaches

The panelists emphasised that solving these challenges requires sustained collaboration between platforms, standards bodies, regulators, and civil society organisations. Success depends not just on technical innovation but on creating governance structures and user experiences that serve diverse communities while maintaining the openness and user agency that make federation attractive.

As Mallory Knodel noted with humour, the discussion completed her “bingo card” of IGF topics, reflecting how federated social media intersects with many core internet governance challenges that the community continues to address.

Mallory Knodel: Hi, everyone. Welcome to the session. I want to first start with a short bit of housekeeping, which is that if you’ve joined us in person, we’re so grateful, and we also have loads of seats up here at the roundtable that we would invite you to occupy if you’d like to. It means you get a microphone that’s always on, so you could ask questions directly. Otherwise, we’ll take questions towards the end, and there are mics also out in the audience, so it’s your choice. We’re also monitoring, for those of you who’ve joined online, we’ll be monitoring the questions that you drop in there as well. So welcome. We’re talking today about the Fediverse, also known as the open social web. By design, imagine this as the openness means that there’s an interoperability element that’s very critical, and we are curious in this panel about a lot of things, but particularly about the ways in which privacy can be preserved in these interoperable environments. So you’re all internet experts. You understand interoperability, and you’re here because you’re excited about the web, and we are too. So this is the panel up before you. We don’t have any remote panelists. I’m going to allow the panelists to really present this idea to you in stages. So if you’re not sure what the Fediverse is, if you’re not sure about what privacy and interoperability have to do with each other, I’m confident that the questions that we’ve posed to one another are going to help build that narrative for you, and of course come with your questions. As well, I’m going to ask the panelists, please, the first time they speak, if they can just give a brief intro into where they work, their interest in the topic. And so that way, I don’t have to read bios aloud to you all. But I should start with myself, actually, right? So they’re all switched on all the time. You have to wear a headset, though. OK, OK, that would help. So yeah, I’m Mallory Knodel. I am the executive director, one of the founders of the Social Web Foundation, which sees itself really as a steward of a multipolar Fediverse. And so that does have elements of protocol. There’s the ActivityPub protocol. There are other open protocols that are hopefully, towards the future, interoperable with each other. It also has this very strong element of community building that we all can build and construct the communities we want online, move them around, subscribe, follow one another without having impediments to that. It’s a really great vision. And I think one of the crucial pieces for me, as someone who’s been in this space with you all for many years, is this element of human rights and of we’re building a Fediverse, or we’re building an open internet, but for what? And so human rights and that sort of thing brings me to this work. So that’s me. I wanted to pose the initial question to the panel, where each person can introduce themselves. Just what do you think we need to know as a level set for this issue of privacy interoperability? It can be a short intervention, but just to try to introduce the audience and the participants to what are the edges of this issue, and why are we sort of here today, what do we care about? So keep the very first initial question maybe two minutes, when the rest of our questions will be maybe around more four or five minutes. So Delara, can I start with you, please?

Delara Derakhshani: Yes, absolutely. Thank you all so much for being here. here and for having me. So my name is Dilara Derekshani. I serve as Director of Policy and Partnerships at the Data Transfer Initiative. Our entire mission is to empower users with regard to data portability. I think it’s helpful to give just a little bit of context about what we do and you’ll see how it’s connected to our work in the Fediverse. When we’re talking about data portability in this context, it essentially means at the user’s request, companies should move bulk transfers of user’s personal data from one platform or service to another. Our work at DTI is centered on translating legal principles into real world practice. On the product side, we are building open source tools with private sector companies. And on the policy side where I sit, we work closely with regulators and policymakers to serve both as a resource and help implement portability solutions that are both meaningful and workable in practice. Ultimately, our goal is an ecosystem that promotes user agency, competition and innovation. And I am heartened to see so many folks at the table today because a lot of these issues are absolutely relevant to the Fediverse and looking forward to diving in deeper.

Mallory Knodel: Great, awesome. Let’s go to you next, please.

Melinda Claybaugh: Hi, thanks for having me and it’s a pleasure to be here. I’m Melinda Claybaugh. I’m a Director of AI and Privacy Policy at Meta. And I’m here today, our interest in the Fediverse and interoperability has really crystallized around our product we launched a couple years ago called Threads, which most people are familiar with Facebook, Instagram, WhatsApp, all of those apps. Threads is our newest app and it is…

Ian Brown: I’ve been working on since 2008 with my co-author, who just happens to be sitting over there, Professor Chris Marsden. We wrote a book in 2013 called Regulating Code, where we talked about interoperability as a solution to a number of difficult policy trade-offs and issues. And I’m really delighted to see it so high on the radar of many regulators. I spent a lot of the last five years talking to EU policymakers because they have actually put it into a law called the Digital Markets Act, which passed a couple of years ago. And I’m going to show you later a couple of tables. I’m not going to talk through them I’m not going to do death by PowerPoint. Don’t worry, but just just point us to a lot more information If you want detailed technical background, I’m a technologist I’ve written a lot about it at my website. You can see it on my website Ian Brown tech I should disclose that earlier this year Metta commissioned from me an independent assessment and it was independent of Of certain aspects of how the digital markets interoperability obligation is being applied to iOS and to iPad OS and the privacy and security Implications of that. I’m actually a security and privacy person by background Although I’ve done a lot in competition digital competition law reform the last five years and I should also disclose that I’m about to fingers crossed start consulting for the UK Competition and Markets Authority on precisely this topic

Mallory Knodel: Yeah, so you can see the spread we’ve got an awesome panel I can tell also from the audience We’ve got some really important stakeholders in this room that know a lot about this topic So I’m counting on you to come in later on with your questions and there are mics there There are a whole bunch of mics up here if you want to sit and take one you are welcome But let me get on with it. So in my view, this is an obvious question, right and we see it a lot There’s an intuition that users have so the the same kinds of users that are excited about using open social protocols for You know consuming Content articles sharing pictures and so on are the same users that are going to care a lot about their privacy and so when this comes into play there, you know if we have more of a permissionless ecosystem Where different entities are now not just maybe in business relationships, which is something I think we’re attuned to in the old social media, but in the new social media, it’s more of you know instances Allowing allow listing block listing that sort of thing. I think there’s an intuition that some users have that and Chris Riley. We’re going to start with a question from a questioner. They said they might take those actions at the instance level maybe because of privacy concerns or maybe because they don’t want to be in relationship with or associate with other corners of this very open interoperable ecosystem. So that’s the sort of general reason why I think we’ve gathered today and hopefully we’ve started out strong by explaining why each of us are here and why we care about that. So first to Dilara, so and I have Ian, I want you to respond also to this question, but first, how can interoperable systems respect user privacy and give users control over their data?

Delara Derakhshani: So I’ve thought about this question a lot, and one of the things I want to start with is what I think, what success would look like in an ideal world, and I think at its core it starts at user agency. Users should know what’s being shared, when, with whom. And why, and interoperability shouldn’t necessarily mean default exposure. You know, there are technical design issues, real-time post visibility across services, activity portability, and, you know, social graph portability as well. I mean, I think there’s an education component that could be furthered to the benefit of all in the ecosystem. But maybe most important is, you know, the idea of, you know, what is the best way to share data? What is the best way to share data? But maybe most important is actually a cultural question, a commitment to shared governance and iterative collaboration. Working in the open is a start, but true privacy-respecting interoperability demands an ongoing habit of listening, engaging with standard bodies, with respect and stay within the space of a secure dialogue that reaches out to everyone at public IoT and later implemented in Azure remote control. And these all strike me as sort of some of the ideals that we should have, I will also note there are at its core a set of smart There are some challenges that come with interoperability. I mean, it’s all about freedom of movement, enabling users to take their posts, profiles, and relationships across platforms. It challenges lock-in, a goal the Fedverse aspires to, but some may argue has not fully delivered on yet. And of course, interoperability is not without its challenges. For example, people may be able to want to move from one server to another and bring all of their content and relationships with them. That may not always be possible. I can speak later to some of the solutions that we’re looking at at DTI. And something that you’ve touched on earlier, and I think it’s relevant to human rights. You know, there’s another challenge, and that’s that decentralized systems like the Fedverse can empower users to find and build communities. But inevitably, that introduces friction. Different services have different norms, policies, controls. It can lead to inconsistent experiences. But beyond that, migration is not just about exporting and importing data. It’s about joining new communities. We’ve heard repeatedly in my experience and our work at DTI that for many users, particularly marginalized groups, the safety of the destination matters just as much as the ability to bring the content. So trust-building tools must also not just address privacy, but also moderation, consent, and survivability. And in recognition of time, I’ll stop there. I’m happy to expand further on any of those points down the line.

Mallory Knodel: Yeah, especially if there are questions from the audience about that. Ian, I want you to respond, but because you’re a technologist, I’m going to ask you if you can slightly explain, maybe for folks who don’t know, that like on ActivityPub, for example, if you delete a post, that doesn’t necessarily mean it’s deleted everywhere. And you could maybe try to hook in some of the actual ways in which these… and I’m going to talk about how these things work and the limits there for privacy and so on. But whatever you were going to say, I’ll do that too.

Ian Brown: That’s good. That’s a challenging question, but I like challenging questions. So, let me say, could our technical friends at the back, I wonder if you could put my first image up on screen, the XKCD cartoon. I’ll just leave you to look at it at your convenience. So, interoperability, I think, first of all, it sounds, to people who aren’t deep in the weeds of this, it sounds a very abstract, weird, geeky, techy notion, and it’s not an end in itself. It’s a means to greater competition, greater user choice, diversity of services, ability of people to talk across communities. I love that idea. It helps people join new communities, try out new communities. You can see this. I wonder if we could have an element of audience interactivity, as we always are told to do in universities to make sure all the students are awake. Put your hands up if you’re on Blue Sky. Yeah, a lot of people. I thought in this IGF community that would be the case. Via a bridge. Well, I was going to come to that. That’s good. That’s also good. Who’s also or differently on Mastodon or another part of the Fediverse? And still quite a few people, which, again, I imagine, I’m not going to go and ask people individually. More of the technical community are on Mastodon for various reasons. More of the policy community on Blue Sky. As Mallory said, there are services now called bridges, which let people, I think the opt-in approach is absolutely right. People who want to can choose to say, OK, from my Mastodon account, I want people on Blue Sky to be able to follow me on Blue Sky and to like my posts and to repost them and to reply to me. So I’ll see them on Mastodon and vice versa. Mallory got straight to one of the one of the important technical questions, which I think this XKCD cartoon, which someone in a tech company just sent me yesterday evening, knowing I was talking on this panel today, the details are very important for human rights, for protecting privacy, for enabling freedom of expression and opinion. for making sure people don’t get harassed by being exposed to communities they might not want to. So to give you another example, there are bridges between X, X Twitter, and some of these other services and I quite understand why many people on Blue Sky and Mastodon would not want people on X to be able to interact with them because there’s a lot of harassment that goes on on that service sadly and I deleted my own account a month or two ago. The very specific question Mallory asked, okay you might be on one service, so on Mastodon for example I know because I do it myself, you can set your posts to auto delete after X weeks or months and I do after two months because I want it to be a conversational medium, not a medium of record. I don’t write social media posts with the care that I write academic articles or submissions to parliamentary inquiries and so on. I want to talk to people, I want to share ideas, I want to be able to make mistakes. That’s a key thing that privacy enables. You can try new things out. It’s actually one of the really critical elements of children growing up, those of you who are parents or have nephews, nieces or are familiar with educational psychology, that people can make mistakes. That’s how we learn. You don’t want people to think if I make a mistake it’s going to be imprinted forevermore on my permanent record and employers, universities, governments, when I enter the United States for example, might be checking my social media posts as is the current US government policy. So that very specific question Mallory asked I think is a really good one because I think it gets to the crux of some of these issues. If say Mallory has, let’s use me as the example, so I delete my Mastodon posts after two months. I use a bridge, it’s called Bridgyfed, I love the name, which lets people on BlueSky read my Mastodon posts. BlueSky is not necessarily currently going to, you know, I have a BlueSky account but BlueSky currently does not have an option to auto delete all my posts. that I type myself into Blue Sky, nor does Bridgyfed currently pick up my preference from Mastodon to auto-delete everything after two months and then apply that on the other side of the bridge. It could, from a technical perspective it’s relatively straightforward. Also from a legal perspective I should mention in Europe at least and in the many other countries, so up to 130 other countries now who have implemented GDPR-like privacy laws and included in the GDPR, I think it’s article 22 but I’d have to check from memory, GDPR says if an organisation publishes your information in some way, if it shares it like a bridge, like a social media service and then you withdraw your consent for the use of that data, the company has to stop using it but also the company has to make best efforts to tell other organisations it’s shared your personal data with, that they should stop using it as well and I think that’s a good legal way of dealing with this issue Mallory.

Mallory Knodel: Good, well and so you set us up really nicely. I think the other, before I move on to the GDPR question, I wanted to just say I think our intuition around how social media works is clearly evolving and I think that while we rely a lot in the privacy realm and the security realm for that matter on how users think about their own data, how users think about their own experience when there’s an absence of regulation, I think this is really really critical and I think why we have to have this conversation now because the new social media is introducing so many different dimensions, it’s introducing so many ways to interact. So let’s talk about regulation because even with the existing regulation we have, even in a sort of old social media regime, there’s still these questions. So Melinda, can you tell us about what challenges you face regarding compliance with national and regional privacy regulations like GDPR and how in a decentralized environment how those challenges can either be exacerbated, alleviated, but you know essentially you know how should Fediverse platforms In instances better aligned with privacy laws in general.

Melinda Claybaugh: Yeah, it’s a really great question. And I guess this is a perfect tee up from you before me And so I think you know these cut these fundamental concepts that are enshrined in GDPR and now in many many many other laws around the world are ones that around your your data subject rights your rights to access information that you’ve shared your right to correct and To transfer and to delete I mean these are these have been enshrined for a long time now and people have come to Expect them and rely on them and operate accordingly And that works really well for things like you know if you post something on Instagram, and then you want to delete it. It’s deleted It works less well In this kind of Fediverse concept so if you post something on threads And then you delete it we will send that request along the chain to wherever you know to Mastodon or wherever it may have also flowed But we don’t have a mechanism for ensuring that that that post is deleted Down the chain, and I think this is where it’s so important That we need to take a fresh look at the existing You know data protection regimes not to say that those rights are aren’t important They should they’re important, and they should stay but the implementation I think is where we need to come together also to Dilara’s earlier point come together with some norms and expectations Within industry around how this should work equally. I think there’s a real challenge as this new social media Proliferates, I think you know it’s those in the tech community are already well-versed in in these communities But to your average user who maybe is starting on threads or starting on blue sky or you know wherever people went after Twitter You know they may not understand this network. They may not understand really what it means to be posting something on threads and then have it go to other services. And so that’s where I think the user education piece and really understanding user expectations, importantly, is going to be really critical. So at Threads, what we do is when you’re first onboarded onto Threads, we you know, kind of explain with pictures what the Fediverse is and what it means for what you’re posting and what happens when you delete. And so I think we have to really meet people where they are and understand kind of who’s a power user, who’s a tech user, and then who’s just your average user who maybe is poking around and trying out new things online, which is great. So I think the the collision of the legal regimes is one that needs to be, I guess the collision of the the new social media and the legal regime is one that needs to be worked out in the regulatory space, in the industry space among companies, and then at the user experience level as well.

Mallory Knodel: Absolutely. And thank you. Yeah, thanks for that. Hopefully there are questions from the audience about that. I think one thing I wanted to just point out is one part of how the ecosystem is evolving now, although it could always change because it’s interoperable and open and, you know, who knows what can happen, is that it feels like a lot of there are kind of two ways to get on the Fediverse. Like maybe you join a new platform for the first time because you want to try it out, like maybe Pixel Fed seems cool. And so you’re new there and you get on boarded there or you’re on already a platform that you’ve been on for a while and that platform may decide to start adopting open protocols. And so the onboarding process and that movement process looks different. And that’s kind of the beauty is that it’s it’s very diverse in how people arrive in this space. And we’re still trying to develop and communicate with end users about how that works. I’m going to move over to another regulation that you’ve already mentioned, Ian. This one’s to you. And then, Melinda, I’d love you. to respond as well. So what are the opportunities to influence the Digital Markets Act? I mean, it’s out, it’s baked, but we want to see if there’s a possibility to define or enforce interoperability requirements for the existing social media platforms. That’s, as far as I know, not part of how the DMA is structured now, but there could be some opportunities and it sounds like you’ve been working on that. So tell us.

Ian Brown: I’ve been working endlessly on that for the last five years. I must have bored all my friends to tears by now on that subject. So the DMA has a messaging interoperability requirement. So that currently applies to WhatsApp and Facebook Messenger and actually Meta, and again, to emphasise, we didn’t practise this before, we did not line up questions, we’ve never met before and my assessment ended at the end of February. But actually, I’m not always positive about Meta, but actually what Meta has done on WhatsApp privacy and interoperability is really interesting and great. I think a lot of user research in fine tuning in the same way Melinda has talked about the Threads experience, exactly the same with WhatsApp, and I think the rest of the industry could really, it would be great if Meta shares that as much as possible with other industry players and open source projects, so they can all take advantage of that great research Meta has done in that area. Civil Society groups got this close to also requiring social media, the biggest, so the DMA only applies to the very, very largest companies, which the DMA calls gatekeepers, there are only seven companies it applies to, the obvious, you can probably guess which they are, I can read the list out later if you care, but Meta is one, Apple is another, Microsoft is a third, ByteDance, TikTok is covered. Civil Society got this close to saying social media from the very, very largest companies must also be interoperable in the way that Threads is. I mean, I’m not using, actually, I am using Threads currently, but mainly I follow people on Threads from Mastodon, not from Threads, I follow Jan Le Coon, for example, who’s Meta’s chief AI scientist. because he says very interesting, quite independent-minded stuff. I’m sure Zuckerberg sometimes rolls his eyes at, you know, things like LeCun says, and that’s a good sign of independence, I would say. What can we do? You know, did civil society miss its chance? Because it persuaded the European Parliament, but it couldn’t persuade the member states, and therefore the final legislation includes it for messaging, but not for social networking services. But this is the cliffhanger. Next year, and actually every three years, every three years the European Commission has to review the operation of the legislation. That’s very standard in European law, and they have to very specifically every three years consider should Article 7, which is the messaging interoperability obligation for gatekeepers, be extended to gatekeeper social media services. And they, you know, we can say pretty much what they would be, probably, because you can look at who are the designated gatekeepers. I already said META is one. You know what social media, they’re called online social networking services in the DMA. You know what online social networking services with 45 million users in the EU, 10,000 small business users, those are the two criteria. You could figure that out for yourself if you wanted to. If the DMA, if the Commission recommended this and the European Parliament chose to act on the European Commission’s recommendation, the DMA could be extended, so that would become a legal obligation for these very largest social media companies in the way it already is for messaging. And for those of you who are not European, don’t worry, I can sort of feel you rolling your eyes that, you know, these Europeans are, you know, they love passing their baroque regulatory regimes, but what relevance does this have to us? Actually, like the GDPR, many other countries around the world are already putting these kind of principles into their own national competition laws and related laws. And I think also another trend we see is traditional competition law is very rapidly moving into this space. So though, if you want to look it up, I’ll just say one sentence about this. There was a European Court of Justice decision very recently called Android Auto, which basically says dominant firms must enable interoperability for their services full stop. Otherwise, it’s an abuse of dominance if they don’t. And I think that could have enormous effects around the world.

Mallory Knodel: I mean, you bring up this idea of jurisdiction, right? So Europe might have done GDPR, has the DMA, the DSA, etc. But I can’t even really picture how this would work if it only applied to Europeans, right? Because I mean, it seems to me, and this is maybe Melinda, I’m sure you have a response here, but to add on to it, you know, how could a company like only be interoperable in one jurisdiction rather than like, technically speaking, it would seem to me to be if you’re interoperable in one place, you’d be sort of interoperable for everyone. But anyway, that’s just my small minded thoughts on that. But tell us your reaction.

Melinda Claybaugh: I mean, I don’t have a specific answer to the jurisdiction question. But I mean, I think it does lead to a larger question around what is interoperability? What are we talking about here, right? Because you can talk about, it’s one thing to talk about, you know, text-based posts, kind of all, you know, that services are being built on a certain protocol, and it’s easier to build that from scratch, right? So something like threads and being able to federate your posts, that makes sense. Messaging, you know, is analogous to email, we kind of all understand that you should, you know, there are benefits to being able to send messages in different ways. But it’s very complicated when you start to get into encryption and encrypted messaging, you kind of start to run into the privacy and security issues. The issues then moving to a social media interoperability, writ large, right, of everything that happens in social media, is a whole other can of worms. And so I think that beyond the technical challenges, and I’m not a competition lawyer, so with the caveat, but what are we trying to accomplish here and how would this actually operate? Is this what people want? But we can leave that for another day in European discussions.

Mallory Knodel: Yeah, I do think that’s relevant, like what do people want? Because I mean, one answer to interoperability all the time on the open web is RSS. I mean, that’s a pretty basic way to just like get content out there in a way that can be consumed in any way. But I think what ActivityPub does and what other protocols are aiming to do is make it that you can push content out there, you can consume content, but you can also interact with it, comment on it, share it, re-blog it. So there’s a lot of different and more all the time, right? Wonderful thing about sort of interoperable permissionless ecosystems is it really creates new ways of interacting with content that didn’t even previously exist before. I mean, we kind of are stuck a little bit in an old school social media thing where it’s like you can click the heart, you can click the little swirly arrow thing, and you can like click the back arrow to make a comment. I mean, there’s a really, really basic ways. There are probably more, right? So Dilara, I want to round out the panel by asking you about, you know, how governance and responsibilities can be distributed so that we can steward all of this together. We’ve talked about this like very diverse and very complicated ecosystem where you can do all manner of things, but then how do we manage this together as it hopefully continues to proliferate, right? Across both existing platforms out there, right? If you run a social media platform and you’re interested or curious about interoperable platforms or interoperable protocols, we want to talk to you, but then also about brand new folks who come into this space and do things. So tell us your thoughts on that.

Delara Derakhshani: Absolutely. You know, earlier I set out a number of challenges and then I didn’t actually bother to explain any of the solutions that we were working on at DTI to address these problems. So I’m going to do that, but you know, I mean, the first thing I’ll say is that it’s pretty clear to all of us that these issues are bigger than just a single company or entity. So you know, federated ecosystems are by design distributed, but that distribution inherently obviously comes with fragmentation. We’ve talked about this already. But what DTI is doing is we’re focused on creating a sort of a shared governance infrastructure, not to centralize control, but to really provide coordination mechanisms that align responsibilities across diverse players. We’ve done this in two ways. The first is we’ve developed a trust model informed by real world use cases and with the input of a great deal of multistakeholder actors on how to establish trust during data transfers. And so, you know, this issue goes beyond just the Fediverse. We see the parallels in the broader online environment. But as I think someone referenced, you know, the Digital Markets Act, which mandates user initiated data transfers for those seven designated gatekeepers, which we can all rattle off who quickly off the top of our head. But you know, it’s silent about trust building mechanisms, and it leaves it up to each gatekeeper. And so to sort of reduce duplication and streamline onboarding across platforms, we’ve launched a trust registry now in pilot. And it allows developers to complete a verification process that, you know, doesn’t have to be duplicated and leads to harmonization and efficiency. There’s also growing recognition that trust infrastructure must scale down, not just up. , we have to recognize that they don’t always have the same resources. And then I think the only other thing I’d like to point out is that with that problem of, you know, I think it would be a good thing to improve portability. And one way we’re doing that is through initiative that we’re calling Lola, which will help users migrate safely. We’re creating shared guardrails and our work is about turning interoperability into a system that users can trust, because otherwise they won’t use it. And where community control and privacy travel together. And I think that’s a good place to stop. At the beginning of a conversation.

Mallory Knodel: Exactly. I like the idea that community control and privacy travel together. I think that’s a good place to stop. I think in a lot, you’ll hear, and I already have so far, like many of you, heard a lot of discussion still about content moderation, social media issues, and so on. And I can’t help but always think, like, right now that we have to sort of advocate and lobby, as human rights activists have done this for many years, just a few companies, right? If you just get like a few of them to like get on board with whatever current issue there is, then you can solve the problem. And in the world that we’re talking about and envisioning and hoping comes to pass, you will have thousands. And so how can you possibly imagine governing or dealing with that complex ecosystem? I think it’s exactly what you’re saying, Dilara. It’s like we now can create efficiencies actually and cross-platform institutions that have the trust that you need that then can hook in and interoperate across the whole ecosystem. So it isn’t about one platform making a decision. It’s about how have we all made the decision and how can it easily replicate across.

Melinda Claybaugh: Exactly.

Ian Brown: Just for 30 seconds, could our technical support put my second image up on screen? I’m not going to talk through it, I’m just going to point it out if you’re interested. You can see I’m not going to talk through it, but this image, and let me just flick to the third, all I want to do is make you aware they exist in the report, the independent report that Meta commissioned, which is on my website, ianbrown.tech. If you want to look at the detail, interoperability, as XKCD and many other people said, you need to get the details right, number one. Number two, it’s great that private firms like Meta and associations like DTI are doing all this research already to figure out, as Mallory said, in a shared space, and I’m very much looking forward to all your contributions the rest of this workshop, figuring out what can we learn together. I promise you, because I’ve done a lot of work for data protection authorities in Europe and the European Commission as well, regulators don’t want to regulate. They don’t have the resources to regulate at this level of complexity and get it right very often. That’s what regulation is for, to give incentives to private firms to do a good job themselves, and as Chris puts it, co-regulation is great, where you can have the private firms and civil society do the detailed work, and the regulators are only there as backstops to make sure that the public interest as well as the private interest is being genuinely represented. The reason these two slides are too complicated, you don’t have to read them now. If you’re interested, download the report, the end of the report. I’m just going to flick backwards and forwards, because they took me so long to do. I want you to enjoy the complexity of these two slides, because this is the level of complexity you have to go into when regulators do have to step in, and regulators don’t want to do this very often.

Mallory Knodel: It’s very instructive, actually. It’s impossible to read, which is maybe what you’re illustrating. Right, so I, as a moderator, am going to pat myself on the back for giving us 20… We have 51 whole minutes for the Q&A. We are not cutting this short because I really, really do want to hear from the audience. I want your questions. We would love to be able to respond. So you can make your way to the mics if you’re sitting out there, and I’ll just try to figure out a cue. If you’re up here already, your mic is actually on. Maybe I should have told you that. Your mic is on, and you can just, like, raise your hand and lean in and get yourself on the list. Yeah, please. If you could introduce yourself, that would be helpful.

Audience: Edmund Chung from .Asia. Very interesting panel. Thank you for bringing the topic in. I have actually two or three questions. One on more social political side, and one on more of the technical side. The first one, well, first of all, I just find the fire chat with the celebrity star Joseph very interesting, because he emphasized on your digital you belongs to you, and this is exactly what we want to talk about. So my first question is, especially to Ian, probably, besides privacy as one aspect, what about the ability to choose, you know, a choice for curation of information, and how that relates to our democratic institutions and processes and so on, because that is a big topic in my mind. In your research, does that, you know, does it help? This is my first question. The second question, related to the deletion part, that’s really interesting. I was just thinking about it. Even email doesn’t resolve that perfectly yet. But something like time-based encryption might be useful. But the issue here seems to be the forum or the standards body that the different players and providers are willing to and Chris Riley. So, it’s a question of whether or not we have a forum that we can abide by. ActivityPub is I think with W3C. Is that going forward? Is W3C the right place? Is IETF, is it a forum, does it need a forum to allow interoperability to really happen?

Mallory Knodel: ≫ I mean, maybe I can answer in reverse.

Ian Brown: ≫ I was going to say you’re more of an expert on question 1.

Mallory Knodel: ≫ Yeah, I think it’s a good question.

Ian Brown: I think it’s a good question.

Mallory Knodel: I think we’re still continuing to standardize extensions, which is exactly what you need to build on this as it grows. So, that still happens, and I think you’re right to point it out as an important piece. As far as like how you ensure different platforms are implementing it correctly and doing things like delete, you know, if requested, that’s something that standards bodies have to do. And I think that’s something that’s important to address, and I think that’s something that’s important to address.

Ian Brown: ≫ And on that issue, I would say the IETF, the internet engineering task force, which I’m much more familiar with than W3C, certainly organizes, I forget what they call them, bake-offs maybe, interoperability testing, making sure that software that is following IETF standards, first of all, deletes the part of the database where misuse happens, so you can counteract corruption at from of an explosive point of view, can drop a number of accounts on that sort that is very potentially a criminal life letter, right? Those are some of the ways that we’re somehow providing sense to other platforms. right approach to sharing their content with other platforms, but then one of those platforms went rogue, you know, started allowing people on that platform to harass them. Well, META can block the harassment, you know, crossing over onto threads, but it can’t enforce against a bad faith actor. Technically, the data still has crossed, you know, it’s on the other side. What could happen then, if necessary, if this was happening on a large scale in particular, META and all the individuals affected could actually take legal action in the 160 countries, including the EU member states that have GDPR-like privacy laws to say, hey, bad faith actor, you’re clearly ignoring my clearly stated limits on processing of my personal data. It’s often going to be sensitive personal data in GDPR terms about your political opinions, your health, a range of other very sensitive topics. And then if your data protection regulators are doing their job, or if your courts are doing their job, they can legally go after the bad faith actors.

Delara Derakhshani: I’d love to quickly respond to something as well, if that’s okay. Your digital you belongs to you. I think that’s an incredibly powerful statement, especially as increasingly our lives are online or increasingly personalization of, for example, AI systems is driving our lives. But I do want to just note that this is the very heart of why portability matters and why you should have control of your data. And with regard to your deletion point, for whatever it’s worth, those conversations are constantly happening, both at the state and increasingly at the federal level. And so I’m looking forward to seeing where those conversations go as well.

Audience: Just quickly, you need to attribute to Joseph Gordon-Levitt, not me.

Delara Derakhshani: I’m so upset that I missed him. I didn’t know he was here.

Ian Brown: He was tremendous.

Mallory Knodel: Yeah, very focused on personal data. It’s really right up your street, Dilara. But it’s recorded, no doubt. You can catch it. We have a question here. Anyone else at the table? Okay, we’ll come to you next. Go ahead.

Audience: Dominique Zelmercier, WCC. So just to quickly react on the interop question, the same way the ITF runs interop tests, WCC has also a very thorough interoperability testing process as part of our recommendation process, which could include indeed testing whether the implementation, implement deletion, whether the service does, is a separate question. I guess a specific question to the panel. So my experience from previous privacy interoperability work in WCC is that it’s really, really hard to define privacy to a level of granularity that matches the complexity of how people want their digital selves to be presented. And so what we’ve seen overall is that what gets through the standardization process is very simple signals like do or do not typically. In a world of social media, that feels way too coarse to actually express something users care about. So I don’t know how you see disentangling this problem of managing the complexity of the way people understand their social self and the need to bring interoperability in exchanging this social media content, in particular in the context of portability or other type of interoperability.

Mallory Knodel: Yeah, good question. Would you like to respond to it? Okay, after. Yeah, okay. Anyone else on the panel want to respond?

Ian Brown: I’m happy to, yeah.

Mallory Knodel: Go ahead.

Ian Brown: That’s a really great point, an important point. And I think one response to that is to emphasize what Mallory and our other two speakers already said about the importance of shared governance and building these norms together collectively, because that’s going to be much more effective than and Chris Riley. I’m going to start with you, Chris. I think the biggest challenge is that we’re seeing big platforms building 10 different versions. With the best will in the world and with very large resources as big platforms have, if they can work together, that’s going to be much better for their users than if they define them slightly differently because that’s going to confuse users. It’s going to lead to things happening that users don’t want because they might have set a privacy control based on their understanding of one thing, but they don’t want to do it because they don’t want to be a part of it.

Mallory Knodel: I think that’s a big challenge. I keep thinking of this instance layer. In the old social media, you have platform, you have users. But now we have users, instances, platforms, and multiple different iterations thereof. So this idea that the instance level can also be a place where default settings are set, where actions can be taken. I think that’s a big challenge. I think there’s a lot of work to do to make sure that your instance could choose to de-fed rate with your instance if you’re not acting in good faith, if you’re not respecting what my users want, if my users actually just want me to do that. There’s a good report that Samantha Lye and Yale Roth at the Carnegie Endowment for International Peace put out about de-fed ration and the politics of that. What does that really look like? I think that’s a big challenge.

Melinda Claybaugh: I think there’s a lot of work to do to make sure that we can customize some of the experiences that we think people will want or make as similar as possible across the federated universe. At the same time, we want to preserve the uniqueness of each of the communities. I think you made a point about algorithms earlier. Maybe you post something on threads and it goes elsewhere and it’s going to be surfaced differently or maybe not like what we were talking earlier. We take over once each community says no than it will harm you. That’s the beauty of it, in a way. You know that you have different types of community armies that can do that really. It’s harming benefits and that kind of thing. I think most of us would think that that’s not so. on different services. So it’s really this fine balance of what are the core things that we want to make sure are protected and actioned across the Fediverse, but what can we leave to to be variable and unique?

Mallory Knodel: Yeah, that’s great. Question down here. Go ahead and introduce yourself and please ask.

Audience: Hi, my name is Caspian. So what are your visions for a privacy-preserving, interoperable internet in 10 years? And what steps should we take to ensure that it exists? Maybe the ideas in 10 years.

Mallory Knodel: I love the idea. We can think backwards to what 10 years ago was like. Very proto of what we have now. And yeah, what’s next in 10 years? I think it’s a good question. Thank you for asking it. Anybody have a response?

Ian Brown: I do. If the other two don’t want to go first. Okay. I was reminded walking in the wonderful display at the entrance by the organizers of how Norway was a real pioneer in bringing the internet to Europe. And I know that in particular because I was looking for a photo of my PhD supervisor who was part of that process. Actually, he was in London and he worked with the Norwegians who are in those photos at the entrance to bring the internet from the United States. I think there were only four sites in the original ARPANET and it was very soon after that that Norway and my supervisor Peter Kirstein brought that over to Europe. So that’s going back to the 70s. So that’s a very long, that’s 50 years ago. I’m not going to even try and remember 10 years ago because it seems all a blur, but thinking 10 years forward is a great horizon. I would say I’m very optimistic what’s happening really quickly. So way before 10 years, actually, I think what’s happening with several of the projects we’ve already talked about today, with Free Our Feets, and I see Robin at the front there, is another great example of stuff that’s bringing this into reality much faster than I expected. software tools. So I’d recommend to you just because I use them personally. I have no financial interest to be clear. One is called Open Vibe that lets you, it’s a social media platform that lets you look at your Mastodon, Blue Sky threads and Nostra accounts all in one timeline. You can customize the timeline. You can adopt different recommendation algorithms as Blue Sky also lets you do. So that point is a really important one Melinda raised. We don’t want to standardize everything. That’s often a criticism of interoperability. You’re standardizing and therefore homogenizing everything. And absolutely we want to avoid that. We want to leave space for different services to compete and develop and do a better job for their users. And that includes the type of communities that they focus on, but we want to give users choice. So if a user on a service, you know, whether it’s threads, Blue Sky, Mastodon, any of these services want to say, look, I’m saying, because a lot of people on Mastodon, for example, are very, very privacy protective for a very good reason. And nothing is requiring them to open up even beyond their own instance if they choose to. And as Mallory said, defederation is the sort of nuclear option if that’s not working properly. So I think one of the things are happening really quickly. I think things are moving so fast here that it’s very, you know, I’ve done these kind of futures exercises before for the European Commission and things are moving so fast. I’m not sure what they will look like in 10 years. As a super nerd, I would say I hope social media looks like the internet at the IP layer, that it’s that configurable and flexible, you know, that it provides a layer in the middle. And then all of the stuff we’ve talked about, you know, the famous hourglass model going up to the, you know, the extra two layers at the top, the ITF t-shirt famously says financial and political, you know, all that stuff is being coordinated by some of these organizations and other people in the room, legal interoperability. So Luca Belli, my colleague from Fundus Audio Tullio Vargas is in the front and has done a lot of work on that. You need all this stuff working together. So this wasn’t this, I didn’t intend this to be a plea for the IGF to continue, but I know it’s been. potentially may not, and I think this is an example of how the IGF can add a lot of value to these debates.

Mallory Knodel: Well, you helped me complete my bingo cards. No, that’s great. And Chris, you had also a question I want to get in. We still have 10 minutes left. So for folks, I see a couple more questions that we should be able to take those as well. Go ahead.

Chris Riley: But the danger of putting the prof in front of a microphone that he didn’t know was already on. So thank you for that. And fantastic panel. Thank you for organizing it as well. I think that often when you look at this question of trying to predict forward 10 years, you also have to look back. And Ian and I worked on the Towards a Future Internet project 16 or so years ago. I remember teaching about interoperability to Melinda’s colleague, Marcus Reinisch, back at Warwick in the last century, the last millennium. And I think that one of the really important lessons that we learned back there, and to mention somebody who probably hasn’t been mentioned for a few IGFs, Lawrence Lessig talked about machine-readable code, lawyer-readable code, and human-readable code. And I had a great conversation last night. One of the reasons why IGF should continue is that you can have great conversations with government officials and engineers and lawyers and policymakers about these things. And one of the fascinating conversations was engineers who had started studying law said, wow, the legal elements of this are really quite straightforward by comparison with what we do. The DMA is very straightforward compared to what we do when we’re working on standards committees. So we don’t find this difficult at all. But the difficulty, of course, is users simply go for defaults. And so one thing I would hope would happen in 10 years’ time, going to Ian’s straw poll of the audience, is that people would find some kind of form of Fediverse that they are comfortable with using in a way where the interoperability is invisible to the user. So Ian and I often cited in the regulating code, but we cited the wonderful Paul Ohm article, the myth of the superuser, the danger of giving people not too many choices, but too many defaults which they find difficult to actually make into a privacy-preserving element. So I guess my question is for everybody, given the last year has been a great experiment in people quiet quitting Mastodon, having noisy quitted X, they’ve quiet quit Mastodon and moved to Blue Sky and a couple of other things, and thank you Ian for turning me on to OpenVibe and learning how I could actually quiet quit in a much more usable way, is how do people see the user out there who doesn’t have, you know, machine readable or lawyer readable abilities to understand these things, how do we make it easier for them to avoid the same problems which occurred, you know, we’re in the middle of a national, a natural experiment, right, where X has become something that almost everybody wants to avoid, except seemingly government official pronouncements which is a very weird thing. So making it more usable for the user, so that in 10 years time users aren’t stuck in this position of having to do more research than they ever want to do, and if I could just mention one thing, I’m now based in Melbourne, Australia, the Australian Competition and Consumer Commission has just issued their final digital platforms report, it’s their 10th over six years, so that has some ideas in it as well I hope.

Mallory Knodel: Yeah, good question, I mean for me I would answer kind of both these questions with, I don’t want to have to sign up for another service ever again, and thank you, I just want to be able to follow all the new people that that service has brought to the internet. But in the interest of time, if I could just maybe suggest that we take, I think we can only maybe take three of these questions, because we have five minutes if they’re quick, and then we’ll see if Dilara, Melinda and Erin have closing remarks, if that works.

Ian Brown: Can you do four, because there are only four people at the moment.

Mallory Knodel: Okay, let’s do, if you can all promise to make it really succinct, like we think we can manage four questions, three responses in five minutes, that math totally works out.

Audience: Yeah, go ahead. So hello, I’m Winston Xu, and I’m from One Power Foundation in Hong Kong, and I’m here, I want to ask a question as now that I’m Many users of the internet are like children. So is there anything that youth can do in the interoperability and fairness of the internet?

Mallory Knodel: Awesome question. Thank you, Wilson. Go ahead.

Audience: I wanted to ask a bit of an open-ended question about more the cultures of each platform and how taking things into a different platform may change the context and remove maybe some the culture from the first platform. I’m thinking of maybe like link sharing and screenshot sharing and how that can very easily spiral out of control. Yeah, how can that be done ethically in a interoperable system?

Mallory Knodel: That’s great. I like that one, too. Thank you for bringing it up. And back over to this side.

Audience: Hello, my name is Gabriel. So the Fediverse being decentralized, you have to deal with lots of different instances, probably hundreds. How do you ensure trust between them? I read a while ago it was found out that, for example, direct messages through Mastodon could be read through if someone was misusing the protocol. So how do you ensure that other instances that are interfaced with can be trusted? Thank you for bringing up direct messages. That’s a good one, especially for a privacy panel. Awesome. And then our last question, please. Good afternoon. Thank you. Perhaps as a question, a reflection, what takes a foot off the toes from the private sector that hopefully, now we talk about interoperability, can be shared? Resilience is about sharing of learnings of what went wrong or what went right. What will you share to the public sector? For instance, in public services, enhancing the services for portable health or for the municipalities, for the social welfare of children or families. What can be shared? For both the good and bad, I’m not focusing only on standards because they are waiting on standards only.

Ian Brown: Thank you. Thanks so much. Let’s go in this order. Ian, if you want to start, just respond to any of that that you can and then we’ll work our way down. Thank you. Incredibly briefly, and let’s keep the conversation going online afterwards. Let me pick out two, the great one from my right, your left, and the last question. So, I think mentioning uncontrolled sharing of screenshots, I think that’s a great way to think about how ActivityPub and at Proto, Blue Skies Protocol and other protocols like it, can help do controlled sharing. So, making sure opt-in, starting point, protocol features that make sure that users’ wishes as far as good faith actors interpret them technically are followed and legal backstops for bad faith actors that choose not to. So, uncontrolled sharing via screenshots is something that’s never going to go away because you can’t stop people screenshotting. That was the dream of the digital rights people in the 90s and the early 2000s. And from a technical perspective, I had a start-up doing it. I promise you, it doesn’t work. That’s a bad idea. Mallory’s way is the way. Mallory and Friends is the way. And the very last question, very simple point, and actually picks up what Chris said as well. One thing governments can do is, for goodness sake, share all your information on multiple platforms, not only on X. Don’t have politicians rail on the one hand about X and then force all your citizens to join X if they want to get information from the government.

Mallory Knodel: Hey, maybe governments should have instances where they’re the source of truth that’s verified.

Melinda Claybaugh: I think what we see is that we’re at the very beginning of this journey. And so I would just encourage, you know, meta is learning as we go. We just last week announced an in-threads feed that pulls, you know, posts from whatever else you’re connected to. That was at the response of users who wanted that. So you know, keep pushing for what you want in this Fediverse, and that will help drive the conversation.

Delara Derakhshani: Yeah, and I’ll finish by saying that a lot of DTI’s work is focused on this. I would really encourage you all to reach out to continue these conversations. The issue of trust during transfers, if a user wants to move their account or content from one service to another, how do we know that the service will respect expectations or that it was actually authorized? How do we confirm identity and integrity and making sure that the message was not lost or maliciously changed? But crucially, you know, these are things that we’re working on, and I really hope that you all reach out for further conversations.

Mallory Knodel: Thanks so much for coming. Thank you for your questions. Thank you to the panelists who spent their time thinking about these issues in depth at the Social Web Foundation and with all of you, right? We are actually really trying to have these conversations together. Like Melinda said, this is only the beginning, so hopefully we can stay engaged. Hopefully we can share learnings and make this space more robust and better. Great.

Agreement points

User agency and control over data sharing

Speakers

– Delara Derakhshani
– Melinda Claybaugh
– Ian Brown

Arguments

User agency requires knowing what’s being shared, when, with whom, and why – interoperability shouldn’t mean default exposure

Users should have rights to access, correct, transfer, and delete their data, but implementation in decentralized systems is challenging

Bridges between platforms require opt-in approaches to respect user consent and prevent unwanted exposure

Summary

All speakers agree that users must have meaningful control over their data and how it’s shared across federated systems, with opt-in mechanisms being preferred over default exposure

Topics

Human rights | Infrastructure

Need for user education and clear communication

Speakers

– Delara Derakhshani
– Melinda Claybaugh
– Chris Riley

Arguments

Privacy-respecting interoperability demands ongoing collaboration with standards bodies and shared governance

Users need clear understanding of what it means to post content that may be shared across multiple services

The goal should be making interoperability invisible to users while preserving choice and community uniqueness

Summary

Speakers consensus that users need better education about federated systems while the complexity should be hidden through good design and clear communication

Topics

Sociocultural | Human rights

Technical challenges require collaborative solutions

Speakers

– Delara Derakhshani
– Melinda Claybaugh
– Ian Brown

Arguments

Federated ecosystems require coordination mechanisms that align responsibilities across diverse players without centralizing control

Existing data protection regimes need fresh implementation approaches for decentralized systems

Different platforms have varying capabilities for features like auto-deletion, creating inconsistent user experiences

Summary

All speakers acknowledge that technical challenges in federated systems require new collaborative approaches and cannot be solved by individual platforms alone

Topics

Infrastructure | Legal and regulatory

Similar viewpoints

Both speakers see the need for institutional mechanisms (trust registries and legal frameworks) to create accountability and efficiency in federated systems

Speakers

– Delara Derakhshani
– Ian Brown

Arguments

Trust registries and verification processes can reduce duplication and streamline onboarding across platforms

GDPR and similar privacy laws provide legal backstops against bad faith actors who ignore user data preferences

Topics

Legal and regulatory | Infrastructure

Both speakers emphasize the importance of designing systems that accommodate different user sophistication levels and allow for experimentation without permanent consequences

Speakers

– Melinda Claybaugh
– Ian Brown

Arguments

Different user types (power users vs. average users) require different levels of explanation and control options

Technical design must enable user mistakes and experimentation while protecting privacy, such as auto-deletion features

Topics

Human rights | Sociocultural

Both speakers recognize that federated systems involve complex community dynamics and governance structures beyond simple technical interoperability

Speakers

– Delara Derakhshani
– Mallory Knodel

Arguments

Migration between platforms involves not just data transfer but joining new communities with different safety considerations

Instance-level governance provides a middle layer between platforms and users for setting defaults and managing relationships

Topics

Sociocultural | Legal and regulatory

Unexpected consensus

Legal frameworks as enablers rather than barriers

Speakers

– Ian Brown
– Melinda Claybaugh
– Delara Derakhshani

Arguments

GDPR and similar privacy laws provide legal backstops against bad faith actors who ignore user data preferences

Existing data protection regimes need fresh implementation approaches for decentralized systems

Trust registries and verification processes can reduce duplication and streamline onboarding across platforms

Explanation

Despite representing different perspectives (academic, industry, and advocacy), all speakers view existing and emerging legal frameworks as supportive of interoperability goals rather than obstacles, which is unexpected given typical industry-regulation tensions

Topics

Legal and regulatory | Human rights

Complexity should be hidden from users

Speakers

– Chris Riley
– Melinda Claybaugh
– Ian Brown

Arguments

The goal should be making interoperability invisible to users while preserving choice and community uniqueness

Different user types (power users vs. average users) require different levels of explanation and control options

The vision includes users never having to sign up for new services while still being able to follow people across platforms

Explanation

There’s unexpected consensus that despite the technical complexity of federated systems, the goal should be to make interoperability transparent to users rather than educating them about technical details, which contrasts with typical tech community emphasis on user understanding

Topics

Sociocultural | Infrastructure

Overall assessment

Summary

Strong consensus exists around user agency, the need for collaborative governance, and making complex systems user-friendly. Speakers agree on fundamental principles while acknowledging implementation challenges.

Consensus level

High level of consensus on principles with constructive disagreement on implementation details. This suggests a mature field where stakeholders share common goals but are working through practical challenges, which bodes well for collaborative solutions in federated social media development.

Different viewpoints

Implementation approach for privacy rights in federated systems

Speakers

– Melinda Claybaugh
– Delara Derakhshani

Arguments

Existing data protection regimes need fresh implementation approaches for decentralized systems

Privacy-respecting interoperability demands ongoing collaboration with standards bodies and shared governance

Summary

Claybaugh focuses on adapting existing legal frameworks like GDPR for federated environments, while Derakhshani emphasizes building new collaborative governance structures and standards-based approaches from the ground up.

Topics

Legal and regulatory | Human rights

Scope and complexity of interoperability requirements

Speakers

– Ian Brown
– Melinda Claybaugh

Arguments

The Digital Markets Act includes messaging interoperability requirements for gatekeepers, with potential extension to social media services under review

Users should have rights to access, correct, transfer, and delete their data, but implementation in decentralized systems is challenging

Summary

Brown advocates for expanding regulatory interoperability requirements to social media platforms, while Claybaugh emphasizes the technical challenges and complexity of implementing such requirements, particularly questioning what interoperability should encompass beyond basic messaging.

Topics

Legal and regulatory | Infrastructure

Balance between user control and system complexity

Speakers

– Audience
– Ian Brown

Arguments

The complexity of privacy controls often results in overly simple ‘do or do not’ options that don’t match user needs

Technical design must enable user mistakes and experimentation while protecting privacy, such as auto-deletion features

Summary

The audience member argues that current privacy controls are too simplistic for complex social interactions, while Brown advocates for features like auto-deletion that prioritize user experimentation over granular control.

Topics

Human rights | Infrastructure

Unexpected differences

Role of government communication in federated systems

Speakers

– Ian Brown

Arguments

Governments should share information across multiple platforms rather than forcing citizens to use single platforms

Explanation

This was an unexpected policy recommendation that emerged during the discussion, with Brown advocating for multi-platform government communication strategies. No other speakers directly addressed this issue, making it a unique position that wasn’t debated but represents a significant policy implication.

Topics

Legal and regulatory | Human rights

Technical impossibility of preventing screenshot sharing

Speakers

– Ian Brown

Arguments

The focus should be on building tools that allow controlled sharing while preventing uncontrolled screenshot-based sharing

Explanation

Brown’s assertion that preventing screenshot sharing is technically impossible and that efforts should focus on controlled sharing mechanisms was unexpected and not challenged by other speakers, despite its significant implications for privacy protection strategies.

Topics

Infrastructure | Human rights

Overall assessment

Summary

The main areas of disagreement centered on implementation approaches for privacy rights in federated systems, the appropriate scope of regulatory interoperability requirements, and the balance between user control complexity and system usability.

Disagreement level

The level of disagreement was moderate and constructive, with speakers generally sharing similar goals but differing on methods and priorities. The disagreements reflect different professional perspectives (policy, technical, regulatory) rather than fundamental philosophical differences, suggesting that collaborative solutions are achievable through continued dialogue and experimentation.

Partial agreements

Similar viewpoints

Both speakers see the need for institutional mechanisms (trust registries and legal frameworks) to create accountability and efficiency in federated systems

Speakers

– Delara Derakhshani
– Ian Brown

Arguments

Trust registries and verification processes can reduce duplication and streamline onboarding across platforms

GDPR and similar privacy laws provide legal backstops against bad faith actors who ignore user data preferences

Topics

Legal and regulatory | Infrastructure

Both speakers emphasize the importance of designing systems that accommodate different user sophistication levels and allow for experimentation without permanent consequences

Speakers

– Melinda Claybaugh
– Ian Brown

Arguments

Different user types (power users vs. average users) require different levels of explanation and control options

Technical design must enable user mistakes and experimentation while protecting privacy, such as auto-deletion features

Topics

Human rights | Sociocultural

Both speakers recognize that federated systems involve complex community dynamics and governance structures beyond simple technical interoperability

Speakers

– Delara Derakhshani
– Mallory Knodel

Arguments

Migration between platforms involves not just data transfer but joining new communities with different safety considerations

Instance-level governance provides a middle layer between platforms and users for setting defaults and managing relationships

Topics

Sociocultural | Legal and regulatory

Key takeaways

Privacy-preserving interoperability requires user agency – users must know what’s being shared, when, with whom, and why, with interoperability not meaning default exposure

Technical challenges exist in federated systems, particularly around data deletion propagation across instances and maintaining consistent user experiences

Existing regulatory frameworks like GDPR provide legal backstops, but need fresh implementation approaches for decentralized systems

The Digital Markets Act’s messaging interoperability requirements may be extended to social media services in upcoming reviews

Shared governance infrastructure is needed to coordinate responsibilities across diverse players without centralizing control

User education is critical as different user types require different levels of explanation and control options

Standards bodies (W3C, IETF) provide essential forums for developing and testing interoperability standards

The future vision involves making interoperability invisible to users while preserving choice and community uniqueness

Resolutions and action items

Continue conversations through the Social Web Foundation and Data Transfer Initiative for ongoing collaboration

Develop trust registries and verification processes to reduce duplication across platforms

Create shared guardrails through initiatives like DTI’s ‘Lola’ project to help users migrate safely

Governments should share information across multiple platforms rather than forcing citizens to use single platforms

Keep pushing for desired features in the Fediverse to drive platform development

Reach out to panelists and organizations for further conversations on trust during transfers and implementation

Unresolved issues

How to effectively propagate deletion requests across all federated instances

Managing the complexity of privacy controls that currently result in overly simple options

Ensuring trust between hundreds of different instances in decentralized systems

Balancing standardization with preserving unique community cultures across platforms

Addressing security vulnerabilities in direct messages and encrypted content in federated environments

Determining what level of interoperability users actually want versus what technologists envision

How to handle cultural context changes when content moves between platforms with different norms

Suggested compromises

Use opt-in approaches for bridges between platforms to respect user consent while enabling interoperability

Implement instance-level governance as a middle layer between platforms and users for managing relationships and defaults

Develop co-regulation approaches where private firms and civil society do detailed work with regulators as backstops

Create shared technical standards while allowing platforms to compete on community focus and user experience

Balance core privacy protections that should be universal across the Fediverse with features that can remain unique to individual communities

Use defederation as a nuclear option mechanism for instances to protect users while maintaining overall system openness

Users should know what’s being shared, when, with whom. And why, and interoperability shouldn’t necessarily mean default exposure… migration is not just about exporting and importing data. It’s about joining new communities… for many users, particularly marginalized groups, the safety of the destination matters just as much as the ability to bring the content.

Speaker

Delara Derakhshani

Reason

This comment reframes interoperability from a purely technical challenge to a human-centered one, highlighting that data portability is fundamentally about community safety and user agency rather than just technical capability. It introduces the crucial insight that marginalized users face unique risks in federated systems.

Impact

This shifted the discussion from technical implementation details to user experience and safety considerations. It established the framework for later discussions about trust, governance, and the need for shared standards that prioritize user protection over technical convenience.

If say Mallory has… I delete my Mastodon posts after two months. I use a bridge… BlueSky currently does not have an option to auto delete all my posts… nor does Bridgyfed currently pick up my preference from Mastodon to auto-delete everything after two months and then apply that on the other side of the bridge.

Speaker

Ian Brown

Reason

This concrete example brilliantly illustrates the complexity of privacy preservation across federated systems. It demonstrates how user privacy preferences don’t automatically translate across platforms, revealing a fundamental gap between user expectations and technical reality.

Impact

This technical example grounded the abstract discussion in reality and led to deeper exploration of GDPR compliance challenges. It prompted Melinda’s response about the collision between new social media and existing legal frameworks, fundamentally shifting the conversation toward regulatory compliance.

I think there’s a real challenge as this new social media proliferates… they may not understand this network. They may not understand really what it means to be posting something on threads and then have it go to other services… we have to really meet people where they are and understand kind of who’s a power user, who’s a tech user, and then who’s just your average user.

Speaker

Melinda Claybaugh

Reason

This comment identifies a critical user experience challenge that could determine the success or failure of federated systems. It acknowledges that most users don’t have the technical literacy to understand the implications of interoperability, which has profound privacy and safety implications.

Impact

This observation redirected the conversation from technical solutions to user education and interface design. It influenced later discussions about defaults, user agency, and the need for systems that work invisibly for non-technical users while preserving privacy.

My experience from previous privacy interoperability work in W3C is that it’s really, really hard to define privacy to a level of granularity that matches the complexity of how people want their digital selves to be presented… what gets through the standardization process is very simple signals like do or do not typically. In a world of social media, that feels way too coarse.

Speaker

Dominique (audience member)

Reason

This comment exposes a fundamental tension between the complexity of human social behavior and the binary nature of technical standards. It challenges the panel to consider whether current approaches to privacy in federated systems are adequate for real human needs.

Impact

This question forced the panel to confront the limitations of technical solutions and led to discussions about shared governance, the role of instances as intermediaries, and the need for more sophisticated approaches to privacy that go beyond simple on/off switches.

The danger of giving people not too many choices, but too many defaults which they find difficult to actually make into a privacy-preserving element… how do we make it easier for them to avoid the same problems which occurred… where X has become something that almost everybody wants to avoid.

Speaker

Chris Riley

Reason

This comment introduces the paradox of choice in privacy settings and connects it to the real-world exodus from Twitter/X. It challenges the assumption that more user control automatically leads to better privacy outcomes, referencing the ‘myth of the superuser.’

Impact

This observation tied together multiple threads of the discussion – user education, defaults, and the practical challenges of federated systems. It prompted reflection on how to make interoperability ‘invisible’ to users while maintaining privacy, influencing the final discussions about usability and adoption.

Overall assessment

These key comments fundamentally shaped the discussion by moving it beyond technical implementation details to address the human and social dimensions of federated systems. Delara’s focus on marginalized users and community safety established a human rights framework that influenced the entire conversation. Ian’s concrete example of auto-deletion across platforms grounded abstract privacy concerns in technical reality, while Melinda’s emphasis on user education highlighted the gap between technical capability and user understanding. The audience questions, particularly about the granularity of privacy controls, challenged the panel to confront the limitations of current approaches. Together, these comments transformed what could have been a purely technical discussion into a nuanced exploration of how to build federated systems that serve real human needs while preserving privacy and safety. The discussion evolved from ‘how do we build interoperable systems?’ to ‘how do we build interoperable systems that people can actually use safely and effectively?’

How can we better educate users about what interoperability means and what happens to their data when they post across federated services?

Speaker

Melinda Claybaugh

Explanation

This is crucial because average users may not understand the network effects of posting on federated platforms and what happens when they delete content across multiple services

How can we develop technical solutions for ensuring deletion requests are honored across the entire federated network?

Speaker

Ian Brown and Mallory Knodel

Explanation

This addresses a fundamental privacy challenge where deleting a post on one platform doesn’t guarantee deletion across all federated instances that received the content

What governance mechanisms can ensure trust and coordination across thousands of diverse federated instances?

Speaker

Delara Derakhshani

Explanation

As the ecosystem scales from a few large platforms to potentially thousands of instances, new coordination mechanisms are needed to maintain trust and shared standards

How can we balance standardization for interoperability with preserving the unique characteristics of different communities and platforms?

Speaker

Melinda Claybaugh and Dominique Zelmercier

Explanation

There’s tension between creating common standards that work across platforms and allowing communities to maintain their distinct cultures and features

How can privacy controls be made granular enough to match the complexity of how people want to present their digital selves?

Speaker

Dominique Zelmercier

Explanation

Current standardization processes tend to produce simple ‘do or do not’ signals, which may be too coarse for complex social media privacy preferences

What can be learned from Meta’s user research on WhatsApp interoperability that could benefit the broader ecosystem?

Speaker

Ian Brown

Explanation

Meta has done significant user research on privacy and interoperability for WhatsApp that could inform best practices across the industry

How can we make federated social media as invisible and user-friendly as email interoperability?

Speaker

Chris Riley

Explanation

The goal is to make interoperability seamless for users who don’t want to become technical experts to use these systems effectively

How can we address the security vulnerabilities in federated systems, such as the ability to read direct messages across instances?

Speaker

Gabriel (audience member)

Explanation

Trust and security between instances is crucial, especially given reports of protocol misuse allowing unauthorized access to private communications

How can context and cultural norms be preserved when content moves between platforms with different communities?

Speaker

Audience member

Explanation

Content sharing across platforms can remove important cultural context and lead to misunderstandings or harassment

What role can young people play in shaping interoperability and internet fairness?

Speaker

Winston Xu

Explanation

Understanding how to engage youth in these technical and policy discussions is important for the future of these systems

What lessons from private sector interoperability work can be applied to public sector services like healthcare and social welfare?

Speaker

Audience member

Explanation

There may be valuable learnings that can improve public services through better data portability and interoperability

How will the European Commission’s review of the Digital Markets Act in the next year affect social media interoperability requirements?

Speaker

Ian Brown

Explanation

The DMA review could extend interoperability requirements from messaging to social media platforms, with global implications

Summary

This roundtable discussion at the Internet Governance Forum 2025 focused on gender mainstreaming in digital connectivity strategies, addressing the significant barriers that prevent women from accessing digital opportunities. Moderated by Rispa Arose from Tenda Community Network in Kenya, the session brought together experts from civil society, regulatory bodies, and community organizations to examine why 2.6 billion people remain offline, with women disproportionately affected by the digital divide.

Mathangi Mohan presented research from the Association for Progressive Communication on integrating gender into community-centered connectivity models, emphasizing that community networks are not automatically inclusive despite being locally driven. The research revealed that women remain underrepresented in community network governance and proposed policy recommendations including gender impact assessments during licensing processes and quotas for women in network leadership roles. Lilian Chamorro from Latin America highlighted specific challenges women face in building connectivity infrastructure, including care work responsibilities, low self-confidence with technology, and limited access to technological devices.

From a regulatory perspective, Dr. Emma Otieno from Kenya and Ivy Tuffuor Hoetu from Ghana emphasized the economic consequences of excluding women from digital strategies. Waqas Hassan shared Pakistan’s success story, where a dedicated national strategy reduced the gender gap from 38% to 25% through whole-of-government and whole-of-society approaches. The discussion revealed that connectivity strategies often focus primarily on infrastructure expansion while assuming equal access, without addressing inherent barriers like affordability, skills gaps, and societal norms.

Key recommendations emerged around the need for intentional measurement and tracking of gender inclusion outcomes, cross-sectoral collaboration beyond ICT ministries, and recognition of digital caretaking as legitimate labor. The panelists stressed that bridging the digital divide requires moving beyond gender-neutral policies to actively address the specific needs and barriers faced by women and marginalized groups in accessing and benefiting from digital connectivity.

Keypoints

## Major Discussion Points:

– **Gender gaps in digital connectivity strategies**: Despite global frameworks like SDG 5 and the Global Digital Compact advocating for digital inclusion, gender is explicitly referenced in only half of national ICT policies, leaving 2.6 billion people offline with women facing disproportionate barriers to accessing digital resources.

– **Community networks as solutions with inherent challenges**: While community-centered connectivity models offer locally-driven alternatives to traditional telecom infrastructure, they are not automatically inclusive – women remain underrepresented in governance due to care work responsibilities, low technical confidence, limited access to devices, and restricted participation in decision-making.

– **Policy implementation gaps versus policy existence**: Many countries have gender-inclusive digital policies on paper but lack intentional implementation, meaningful measurement, and disaggregated data collection to track outcomes and impact, resulting in continued exclusion despite regulatory frameworks.

– **Economic consequences of digital gender exclusion**: The exclusion of women from the digital economy has caused approximately $1 trillion in GDP loss over the last decade in developing countries, with potential for another $500 billion loss over the next five years if current trends continue.

– **Cross-sectoral collaboration as essential solution**: Effective gender mainstreaming requires moving beyond ICT ministry silos to involve education, finance, health, and gender ministries in a “whole of government” approach, supported by multi-stakeholder partnerships including civil society, private sector, and community representatives.

## Overall Purpose:

The discussion aimed to explore how to effectively mainstream gender considerations into digital connectivity strategies, moving from global policy commitments to practical local implementation. The session sought to identify barriers preventing women’s participation in digital infrastructure development and propose actionable solutions for creating more inclusive connectivity models, particularly through community networks.

## Overall Tone:

The discussion maintained a professional, collaborative, and solution-oriented tone throughout. It began with a serious acknowledgment of the scale of digital exclusion challenges, evolved into detailed problem analysis with speakers sharing candid experiences about implementation gaps, and concluded on an optimistic note with concrete recommendations and successful examples like Pakistan’s gender strategy. The tone remained constructive and forward-looking, emphasizing shared responsibility and the urgency of action while celebrating incremental progress and best practices.

Speakers

**Speakers from the provided list:**

– **Rispa Arose** – Works for Tenda Community Network based in Nairobi, Kenya; Session moderator focusing on grassroot-driven development by expanding access to connectivity and digital opportunities

– **Mathangi as Rispur** – Works at the intersection of gender technology and digital governance; leads programs in partnership of Data Currency, Artificial Intelligence; works with Self-Employed Women Association in India leading decentralized livelihood and digital upskilling program for women in the informal economy

– **Lillian Chamorro** – From Colnodo; supports community-centered connectivity in the Latin America region with focus on gender inclusion in grassroots community-centered connectivity initiatives

– **Dr. Emma Otieno** – Digital Inclusivity Champion; serves as Kenya country coordinator for REFEN (Réseau Internationale des Femmes Expertises du Numérique); previously served as Deputy Director for Universal Service Fund and Manager for Strategy Development and Fund Mobilization

– **Waqas Hassan** – Representative from Global Digital Partnership (GDIP); spearheads policy and advocacy engagement for Global Digital Inclusion Partnership in Asia; has background in inclusive policy development and gender responsive strategies

– **Ivy Tuffuor Hoetu** – Senior Manager, Regulatory Administrator with National Communication Authority in Ghana; ICT Development and Digital Inclusion Advocate; has experience in digital connectivity, internet development and ICT training

– **Josephine Miliza** – Works for Association for Progressive Communication as global policy coordinator; provides civil society perspective on integrating and mainstreaming gender in connectivity strategies

– **Kwaku Wanchi** – From Ghana IGF

**Additional speakers:**

– **Dr. Rhonda Zalesny-Green** – Co-founder and director at Panoply Digital; co-authored report on gender integration in policy and regulation for community-centered connectivity models (mentioned but did not speak directly)

# Gender Mainstreaming in Digital Connectivity Strategies: A Comprehensive Roundtable Analysis

## Executive Summary

This roundtable discussion at the Internet Governance Forum 2025 examined gender mainstreaming in digital connectivity strategies, addressing why 2.6 billion people remain offline with women disproportionately affected by the digital divide. Moderated by Rispa Arose from Tenda Community Network in Kenya, the session brought together experts from civil society organisations, regulatory bodies, and community networks across Africa, Asia, and Latin America.

The discussion revealed that whilst global frameworks advocate for digital inclusion, gender is explicitly referenced in only half of national ICT policies. More critically, even where policies exist, they often lack intentional implementation and meaningful measurement mechanisms. Speakers highlighted that gender exclusion from the digital economy has caused approximately $1 trillion in GDP loss over the last decade in developing countries, with potential for another $500 billion loss over the next five years if current trends continue.

## Key Themes and Findings

### The Infrastructure Fallacy in Digital Inclusion

A central theme emerged around the misconception that infrastructure development automatically translates to inclusive access. Waqas Hassan from the Global Digital Partnership explained: “The connectivity strategies are still primarily developed from an infrastructure mindset… that mindset is based on the assumption that if you provide access and expansion, coverage expansion, that is somehow synonymous with inclusion… infrastructure development does not mean equitable access.”

Mathangi Mohan from the Association for Progressive Communication reinforced this point: “If we don’t address or even consider the gendered digital divide, then you’re not really closing the gap, but more like building the infrastructure on top of the inequality and marginalisation that is already existing.”

### Community Networks: Opportunities and Barriers

The discussion examined community-centred connectivity models as potential solutions to traditional telecom limitations. However, speakers emphasized that community networks are not automatically inclusive despite being locally driven. Mathangi Mohan presented research findings showing that women remain underrepresented in community network governance structures due to multiple intersecting barriers.

Lillian Chamorro from Colnodo in Latin America identified specific barriers women face, including care work responsibilities, low self-confidence with technology, and limited access to technological devices. She noted that “women prefer to stay away from technical roles” due to these confidence issues and structural limitations.

### Regulatory Perspectives and Implementation Gaps

Dr. Emma Otieno from Kenya provided a candid assessment of policy failures: “We have not been intentional in terms of putting in these clauses in the policy regulation strategies, we are putting them in as aspects of satisfying the social or compliance aspects, but not meaningfully and intentionally putting them… What has lacked completely is measurement, the intentional and meaningful tracking.”

Ivy Tuffuor Hoetu from Ghana emphasized the need to move “from equality to equitable distribution,” using the metaphor: “You cannot give us the same level of a tree to climb or something. One of them will need a ladder to be able to climb the tree.”

Both regulatory experts identified the lack of gender-disaggregated data as a fundamental barrier to creating targeted policies and noted limited coordination across ministries responsible for gender, education, and health.

### Success Stories

Waqas Hassan shared Pakistan’s achievement in reducing the gender gap from 38% to 25% through a dedicated national gender strategy. This success was attributed to inclusive policymaking processes involving multiple stakeholders and adopting both whole-of-government and whole-of-society approaches, with formal institutional structures including high-level steering committees and official working groups.

## Areas of Consensus

### Intentional Design and Implementation

All speakers agreed that gender mainstreaming requires intentional design from inception rather than treating gender as an afterthought. This consensus emerged across different sectors and regions.

### Cross-Sectoral Collaboration

There was unanimous agreement on the need for whole-of-government approaches extending beyond ICT ministries, requiring coordination with education, finance, health, and gender ministries, supported by multi-stakeholder partnerships.

### Local Context and Cultural Sensitivity

Speakers agreed on the importance of respecting local contexts, languages, and cultural values whilst creating supportive spaces for women to engage with technology, including translating digital skills training into local languages.

### Economic Imperative

Speakers emphasized that gender digital inclusion represents both a social justice issue and a critical economic imperative with measurable financial consequences.

## Concrete Recommendations

### Regulatory and Policy Reforms

– Implement gender impact assessments during licensing processes for community networks

– Create financing funds with special conditions for women’s participation in technology access and training

– Accommodate informal women’s groups and non-registered collectives in licensing frameworks

– Establish meaningful measurement, tracking, and monitoring systems for gender inclusion policies

### Community-Level Interventions

– Create women’s circles for dialogue and expression whilst respecting community values

– Translate digital skills training into local languages using retired language teachers

– Support collaboration between community networks and community libraries or information centres

### Funding and Sustainability

– Create infrastructure funds with contributions from all sectors benefiting from ICTs, not just telecom operators

– Develop financing mechanisms that support women-led community networks at scale

## Outstanding Challenges

The discussion identified several unresolved issues requiring further attention:

– How to effectively redistribute care work responsibilities that prevent women’s participation

– Developing sustainable funding models for women-led connectivity initiatives

– Balancing respect for traditional community values with promoting women’s technical participation

– Creating standardized approaches for collecting gender-disaggregated data across different contexts

## Conclusion

The roundtable demonstrated both the complexity of gender mainstreaming in digital connectivity and the potential for coordinated action. The high level of consensus among speakers from different sectors and regions suggests significant potential for collaborative policy development and implementation.

The session established that bridging the digital divide requires moving beyond gender-neutral policies to actively address specific barriers faced by women and marginalized groups. This represents a fundamental shift from treating digital inclusion as a technical challenge to recognizing it as a complex social and economic issue requiring equity-focused approaches.

The path forward requires sustained commitment to intentional implementation, meaningful measurement, and collaborative action across sectors and stakeholders. As demonstrated by Pakistan’s success, the tools and knowledge exist to address these challenges effectively; what remains is the political will and institutional commitment to transform policy commitments into practical results for women and marginalized communities worldwide.

Rispa Arose: Hello everyone, and a warm welcome to this roundtable discussion. To our online audience, good morning, good evening, good afternoon from where you’re joining us from today. It has been an enriching and insightful week, participating at the Internet Governance Forum 2025 here in Norway. My name is Rispa Arose, I work for Tenda Community Network that is based in Nairobi, Kenya, where we work towards fostering grassroot-driven development by expanding access to connectivity and digital opportunities. We are honored for this opportunity and platform to bring our local voices on this global stage. Today I have the privilege of moderating our discussion on a topic that really lies at the heart of digital inclusion, gender mainstreaming in digital connectivity strategies. Many of you, if not all of you, are aware that we have 2.6 billion people that are still offline. That’s a huge number and that’s a big challenge. As we know, digital connectivity is no longer a luxury, it’s really a critical enabler of socio-economic participation. Despite global efforts to close this digital gender gap, women continue to face significant barriers in accessing financial, educational, social and health resources in this digital age. And while some governments have introduced policies to address this divide, gender mainstreaming in ICT policy and regulation remains limited or sometimes even absent. According to different frameworks, such as, according to the ITU, gender is explicitly referenced in only half of national ICT policies or master plans and the frameworks such as WSIS Action Plus 20, the Global Digital Compact and Sustainable Development Goal SDG No. 5 collectively advocate for an inclusive infrastructure and digital rights. However, without intentional gender mainstreaming in digital strategies, millions of women and girls will remain excluded from the benefits of the digital economy. So this session will explore how global commitments can translate into local actions by implementing gender responsive policies and regulations in the different frameworks as well as look into how to mainstream gender and digital connectivity strategies. I’m thrilled to have a distinguished panel of experts and practitioners with me, with us today, both here in person as well as online. Three are joining online and I’ll begin with the ones we have here. So I’ll start with Josephine Miliza who is here and works for the Association for Progressive Communication as their global policy coordinator and she will be really giving us insights from a civil society perspective on what it takes to integrate and mainstream gender in connectivity strategies. We also have Lilian Chamorro from Colnodo who is also part of supporting community-centered connectivity in the Latin America region and will also be really sharing from for her experience working with this community-centered connectivity, what it means to support women and ensure that there is gender inclusion in grassroots community-centered connectivity initiatives. Online, I’m joined with Dr. Emma Otieno, who is a Digital Inclusivity Champion and serves as a REFEN, which is Réseau Internationale des Femmes Expertises du Numérique, and she is the Kenya country coordinator championing for digital gender inclusivity and promoting women’s leadership in the digital technology space globally. Previously, she has served in other capacities as Deputy Director for the Universal Service Fund and also the Manager for Strategy Development and Fund Mobilization Manager. She will be sharing a lot from a regulator perspective around this topic. Then we have Waqas Hassan, who is a representative from the Global Digital Partnership, the GDIP. He spearheads the policy and advocacy engagement for Global Digital Inclusion Partnership in Asia. He has a strong background in inclusive policy development, especially the gender responsive strategies and program. He has also implemented connectivity and community empowerment projects, with particular focus on underserved areas. Then we have Ivy Tufor Hoytil, who is a Senior Manager, Regulatory Administrator, working with the National Communication Authority in Ghana. At this specific meeting, she will be… presenting in her own capacity as ICT Development and Digital Inclusion Advocate. She has done work, she has considerable experience in digital connectivity, internet development and ICT training as well as also has been responsible for monitoring radio spectrum bands, resolving maintenance frequency database and enforcing authorization conditions. Then last but not least, we have Mathangi Mohan, who works at the intersection of gender technology and digital governance. She currently leads programs in partnership of Data Currency, Artificial Intelligence where she designs global capacity building initiatives on responsible artificial intelligence and data governance across 80 countries. She is working now with the Self-Employed Women Association in India where she leads decentralized livelihood and digital upskilling program for women in the informal economy. So that’s our lineup of speakers and I’d like to get right into the session and now welcome Ms. Mathangi Mohan to give us a keynote presentation on a recent research conducted by the Association for Progressive Communication on integrating gender in policy and regulation for community-centered connectivity models. Ms. Mathangi Mohan, please, you can have the floor.

Mathangi as Rispur: Thank you, thank you, Rispur, and thank you to APC and also the Internet Governance Forum for the opportunity to be a part of this discussion. My name is Mathangi as Rispur just introduced me and I’m speaking on behalf of my colleague Dr. Rhonda Zalesny-Green. co-founder and director at Panoply Digital and she co-authored this report with Shivam and our colleagues, our team on the gender integration in policy and regulation for community-centered connectivity models. I work alongside Ronda at Panoply on gender technology and connectivity models and governance at both a grassroots level and also in the policy spaces. So today I’ll be sharing insights from the report while also drawing connections to the practice and lived experiences from my own experience as well. So we can start with, let me start with the basics. So what are community networks or CNs? While they are not new, but they’ve been gaining global attention for a good reason because these are internet networks that are built and managed by local communities themselves. They usually operate in places where large telecom companies don’t see a lot of commercial value like remote areas, informal settlements or underserved regions. But what makes them so powerful is not that they fill coverage gaps, but they are mostly locally driven and because of their cooperative spirit and also often they are non-profit, they allow communities to control and own their digital infrastructure, deciding how it is built, how it is maintained and who it benefits. In many ways these CNs represent a form of digital sovereignty as well because they can support local businesses, they can strengthen education and healthcare systems and even provide emergency communications when most national systems fail. And because they are rooted in community needs, they can adapt very quickly and responsibly in ways that traditional internet service providers and telecoms often cannot. And that is what our report was trying to lay bare, that these networks are not automatically inclusive just because they operate in this way and just because the network is community run. It does not mean that everyone in the community is equally represented and we have taken a gender angle. Women, especially in rural and low-income contexts, remain underrepresented in these CN governance. They are most often underpaid or most often offers a suite of policy recommendations. I would like to just highlight a few examples because we have tried to approach it from how do we tailor the recommendations for each ministry involved because we wanted to adopt a whole of government approach as well. So, one would be for the ministries of ICT in these countries and beyond should require a gender impact assessment during the licensing processes itself for the CNs or some kind of an incentive for women-led CNs and they can design more simpler license procedures to reduce the barriers for women to access the leadership roles in these cooperatives or in these networks. And when it comes to the gender ministries itself, they can introduce quotas for women in the CN governance, fund women-only training spaces when it comes to community networks or access cooperatives and also campaign against online gender-based violence because that is also a major deterrent which is keeping women offline. And to the finance ministries, they can of course offer subsidies for smartphones or for connectivity for low-income women or provide tax breaks, consider providing tax breaks to the CNs with a strong gender inclusion or women-led CNs. And finally, for education ministries to mandate that public ICT… and all the other researchers. And I just thought it’s really important to highlight what Risper earlier said, that we often want to say that we want to bridge the digital divide. But if we don’t address or even consider the gendered digital divide, then you’re not really closing the gap, but more like building the infrastructure on top of the inequality and marginalization that is already existing. And community networks are, they offer, they give us an opportunity to do things differently, to make things right. And we definitely need to take the opportunity up and design for it and regulate in it and invest more in these networks. And yeah, with that, I want to say thank you, and I look forward to the discussion as well.

Rispa Arose: Thank you. Thank you so much, Ms. Mathangi, for really giving us a grounding on what community networks are, what community networks bring towards bridging the digital divide, as well as speaking to some of the challenges that exist with regards to thinking of connectivity, not thinking of gender, when thinking about connectivity, having gender not as an afterthought, but thinking through it, even at the inception stage, at the design stage. And this is, as you’ve correctly said, is the opportunity that community. Thank you so much for joining us for this session. We’re going to be talking about the challenges that community networks bring in connectivity strategies. Now, this really gives us a good segue to the next speaker, who will touch on some of the challenges that women face when trying to build and sustain this community network infrastructure. So, I’m going to turn it over to you, Lillian, to talk a little bit about the challenges that women face when trying to build this network, but also supported the ecosystem, we’ll share her experience on what are some of the specific challenges that women face when trying to build this infrastructure, this community-centered connectivity infrastructure. Lillian, over to you.

Lillian Chamorro: Thank you so much for having me. I’m going to start with you, Emma, and then I’m going to turn it over to you, Lillian. Okay, from the experience of accompanying this type of initiative, we can mention some of the challenges for women. One of them is the care work that women must assume makes it difficult for them to participate in the different activities involved in the implementation, but also in the maintenance of the network. So, this is a challenge for women, especially in rural areas, where there is a lot of work that is done exclusively in women. Other challenges for some women, especially in rural areas, have low self-confidence, as they do not feel capable of appropriating technologies and technological knowledge, and access to the technological knowledge. So, they prefer to stay away from the internet, but also, in general, the developing countries, there are some and Dr. Josephine Miliza, Dr. Emma Otieno, Waqas Hassan, Ronda Zelezny-Green, Ivy Tuffuor Hoetu Hoetu. Also around the low participation in decision-making and leadership. Then positive actions are required, but also appropriate methodologies that provide women with confidence in approaching and appropriating these issues. All these result in more difficulty to be part of workshops or participate in training processes. And in many occasions, there are difficulties to leave the home and participate in these kind of spaces. Finally, I can mention lower access to technological devices, since other expenses are privileged and there are not appropriate technological tools to assume certain tasks. Thank you.

Rispa Arose: Thank you so much, Lilian, for sharing some of the challenges that women face when trying to build and sustain connectivity infrastructure. Being just working for TANDA community network, I totally relate with some of these challenges, starting from the low confidence level. It really takes a lot, especially because community-centered connectivity is very technical. There has to be some encouragement, some technical training, some sort of mentorship to even get women coming in to support the network infrastructure, be it of the connectivity or the community. This really helps to frame the challenge of this discussion today. What I would like to hear from the next speaker. Waqas, who’s joining us online, is what do you see as the core disconnect? Why gender is still largely missing from connectivity strategies? Why do we still have a huge gender gap within the different connectivity strategies? And yes, over to you, Waqas.

Waqas Hassan: Thank you. Thank you, Christopher. I hope I’m audible. So I think much of what you just asked was covered by Matangi as well. But if you see it from a holistic level, there are, of course, several factors which are dependent on geographical profile and the level of ICT development and things like that, and lack of data awareness capacity of the government itself to understand and then act on the problem. But from my own experience, from my previous life with the government, I think there are a few things that I have observed, and I think two of them stand out for me, which I would like to, of course, share with all of you. The first one is that the connectivity strategies are still primarily developed from an infrastructure mindset. And that mindset is based on the assumption that if you provide access and expansion, coverage expansion, that is somehow synonymous with inclusion. Now, I think that if a particular community is then provided with mobile connectivity or broadband or whatever they mean, there is now an equal opportunity for men and women to use the digital services. But I think what they need to realize is that irrespective of how much technological innovation we bring about, infrastructure development does not mean equitable access. There are inherent barriers of affordability, skills, societal norm, online safety, and security. These are the Thank you so much for joining us for this session of the Global Dialogues. I’m Ronda Zelezny-Green, and I’m here with Dr. Emma Otieno, Dr. Ronda Zelezny-Green, and Dr. Waqas Hassan, Dr. Emma Otieno, and Dr. Ronda Zelezny-Green. Thank you so much for joining us. And there is undeniable evidence around that. Our own research, GDIP’s research, it proves that the exclusion of women from the digital economy in the developing countries has caused about $1 trillion in the loss of GDP over the last decade. And if it continues, another $500 billion could be lost over the next five years. So I think this realization and this advocacy angle that we also pursue as GDIP when we talk to the governments is that it is not just a women empowerment issue or a social issue. It is actually an economic crisis that we want to work here, all of us together. I think I’ll just end with saying that this is the realization that needs to be propagated through our interactions with the governments, that they have to integrate gender in the connectivity policies and then take concrete actions on it, not just mention it like we see in many of the policies like the IT policies or the broadband policies, and there is some way of mentioning those kind of instruments, but never really the essence of those policies. So I’ll just stop there. And yeah, looking forward to the discussion. Thank you.

Rispa Arose: Thank you so much. Thank you so much, Waqas, and thank you for bringing in such heaviness on the consequences or the grief, loss that we face when we are not inclusive in our strategies, when we are, when the agenda is left behind in promoting or in, for example, connectivity strategies, then it means not only social loss, but it goes to even economic loss from a global perspective. With that in mind, I want to move to the next speakers, who are the regulators, Dr. Emma from Kenya and Ms. Ivy from Ghana. From a regulatory perspective, why is it important to integrate gender in digital connectivity strategies? And building on what Waqas has just mentioned, as some of the losses that we are having because of the gap, gender gap, what do you think is at stake if we don’t mainstream gender in these digital connectivity strategies? I’ll start with Dr. Emma, and then we’ll move to Ivy, Ms. Ivy.

Dr. Emma Otieno: Okay, thank you very much, Rispa, and I really want to thank the speakers who have gone before me. Can you hear me? Yes. Okay, so I just want to start by looking at the question you’ve asked, why? Why should we really worry, or why should we really prioritize this subject of agenda inclusivity? And my simple answer would be that we have contracted both at the global level and at national levels, and agreed generally that we will not leave anyone behind. And we said we will not leave anyone behind because we want the world connected to the last person who’s under a cave, or if there’s another one under the sea, they must all be connected. So, for me, when I look, when I sit at the national level, that is the mantra of a digital transformation, leaving no one behind. When we go international, be it at the UN level, be it at the International Telecommunications Union level, be it at any other forums that we and any other place where the subject of digital transformation is being pursued, the agenda has been leaving no one behind. So, when we come with the eye of the regulator, because we’ve been actually been, why regulation exists is to create enabling environments so that the government can achieve the broader social economic development goals. So, for us to achieve that, leaving no one behind, which is very critical for the government’s social economic development, we must pursue goals, we must pursue regulations that

Rispa Arose: Thank you so much, Dr. Emma. Over to you, Ivy.

Ivy Tuffuor Hoetu: Yes, thank you. And am I audible? Yes, we can hear you. Okay, good morning from my location. So one of the blind spots I would touch on is most of the time we perceive technology as gender neutral and policies are made that technology will benefit everyone equally. But we have diverse needs. And so one thing we are lacking or is at our blind spots is equitable access and distribution of technology as against gender neutrality of technology. It is assumed that technology or digital policies are inherently gender neutral and will benefit everyone equally, which is not the case. Most of the time, the focus has been on overall connectivity metrics without disaggregating data or considering how societal gender inequalities translate into data disparities. And this is largely due to data deficiencies in gender disaggregated data on data access, usage patterns, and online experiences. Governments have been promoting gathering of ICT indicators and all that, but the in-country, the national, the countries, how often do we disaggregate this data? Even if we have to analyze them, do we analyze them based on the gender specific needs? And here, gender, I’m talking about going even beyond women. The gender definition cuts across children, the physically challenged, the elderly, and women, which most of these people are mostly underrepresented. And policies or regulations usually does not affect their specific needs. And it’s all because the data we collect or we gather, we collect them holistically without disaggregating them. So policies are made generally instead of maybe equitable basis. And it makes it difficult to identify these specific needs. So this is one of the blind spots that we are lacking. And I believe if it’s addressed holistically, it can address and target the needs of all these specific genders. Most of the time, yes, it’s women, but there are also even the elderly, we are saying inclusivity, connecting everyone everywhere, ISOC goal. But how can we reach out to these people if we don’t understand their specific needs? If we make regulations or laws equally, that whatever we have in place will benefit everyone equally. It cannot. My height, I am short. Someone is shorter or someone is taller. You cannot give us the same level of a tree to climb or something. One of them will need a ladder to be able to climb the tree. The other will not need a ladder. So we need to disaggregate the data we collect, make specific indicators based on the specific gender needs. And I believe it will address these blind spots. Thank you.

Rispa Arose: Thank you so much, Ms. Ivy. Thank you so much, Dr. Emma, for bringing in the regulatory perspective to this conversation. Before we move into the next segment, which we’ll be looking at now, we’ve talked about the challenges, we’ve talked about the problems, the gaps that exist, and what is at stake. Before we move to some of the solutions, I’d like to see if there’s any question or any comment online or in person. If you have any question, you can just move to the roundtable. We can take it. Yes. We can have that question and then move to the next segment.

Kwaku Wanchi: For your presentation, my name is Kwaku Wanchi from Ghana IGF. There’s something so organic of what you’re talking about in terms of community network, which is speaking about collaboration and being able to provide solutions to the problem. One of the things that I always see as being helpful in some of this is more I want to know more about what I would call our community businesses. And again, this is gender-based or women-led, especially in Africa. I don’t know about Lilian’s experience in Latin America and the rest. And what you would find, I think there was a report about a year or two ago from MasterCard Foundation, which found that most businesses in Africa are run by women and small, medium enterprises. That’s what I mean. And I think one of the things and the challenges and opportunities in thinking about how we go, I don’t know how the next conversation is going to be, is one, what I would like to call synthesizing or having organic technologies or solutions in terms of connectivity. Where you’d see that most of our local businesses or small, medium enterprises are small businesses, but how do we get them digital or being able to do it? An aspect about the work we do in community networks is about imparting digital skills. And I always give this keynote, coming from Africa, we are multilingual. And most of the things that we do need to be owned by the people who can be able to speak their own language or interpret it. So my point is this. The skills impartation and things that we do, we need to be able to translate into our local languages. And I’m giving this for free. I’ve been doing this for five years. Let’s try and translate things into our local languages. And the best people to be able to do that are language teachers or retired language teachers. That’s what I would say. That’s inclusivity in gender. Second part is including our medium-sized businesses to get them digital, to teach them the skills. And also give them their business experience and create that business environment which gives opportunities to give value to our community. And you’d be surprised. You probably might be getting financing from some of these local businesses. So that’s what I’d like to contribute. Thank you.

Rispa Arose: Thank you so much for that contribution. I believe it was just to add to the conversation. The speakers agree that diversity should go beyond just the gender, but also look at languages. Look at age. How can different age levels contribute to digital connectivity for different communities? This now will take us to the next question. I don’t know if there’s any question online.

Josephine Miliza: There’s no question, just a comment from Peter Balaba. He says, I do support the establishment of community networks approach in Kenya. However, collaboration with community libraries and information centers can help bridge the gaps. In Keseke Telecenter and Library, we are now focusing on empowering the youth and women in digital skills and we train them in the local language to enable them to understand the basics. So it’s just a comment and not a question. Thank you, Peter.

Rispa Arose: Thank you. Thank you, Josephine. To the next segment, which we will have another round of interventions from our speakers, we’ll be looking at what it will take to realize the gender inclusive connectivity. And now I’d like to go back to our first speaker that gave us the presentation remark, Ms. Mathangi and Lilian, if you can talk about what actions are necessary to better support the emergence and sustainability of women-led connectivity models. You’ve already mentioned what are some of the challenges and barriers for women to participate in community-centered connectivity models. Now, what are some of the actions that can be taken? For Mathangi, you can talk about some of the recommendations that that came out of the research that you’ve just presented. So I’ll start with Ms. Mathangi and then we can go to Lillian.

Mathangi as Rispur: All right, sure, thanks, Rizpal. And while also talking about the policy lens and recommendation from the report, I would also like to share something that I’ve seen in practice. Because while working with some of the self-help groups and particularly the Self-Employed Women’s Association here in India, I’ve seen where we supported some of these artisan-led cooperatives access digital markets, and that too, during the first and the second waves of the pandemic. And then what happened was something very interesting was that to address the connectivity gaps in one cluster, in several clusters of villages, actually, the women managed to have shared hotspot system, not like the community network, but also not very unlike the CNs that we spoke about earlier. And they used to train each other, watch tutorials, and figure out how do they ensure it gets to the online inventory and all of those, even troubleshooting the basic tech issues. They supported each other, but the catch was it was sustainable only because the initiative worked because it was under an umbrella like SEWA where they could step in to subsidize the hardware. They could, like one of our audience members pointed out, they coordinated all the training in their local language, Gujarati, and they anchored the work in existing self-help groups. So then what finding came out was very similar to what we also studied in the report that if you remove that scaffolding, that model would not have survived or it would not have been sustainable. So combining these two insights, just quickly summarizing three recommendations would be, first, the licensing frameworks for these CNs need to very explicitly accommodate non-registered collectives, informal women’s group, like some of them rightly pointed out that the MSMEs and SMEs, these are majorly owned by women. So supporting a more informal women’s group, that is also equally important because these are mostly cooperatives rooted in trust and solidarity and just assuming that it should be a registered entity and a technical documentation. So that’s also going to exclude a lot of groups. Secondly, there are a lot of already funding mechanisms, but one key insight is that the funds should also allow for experimentation and thereby allowing failure and more iteration to figure out what works and what does not work, because these community networks are not going to scale if you take a startup timeline and when we compare. And third, should be echoing Waqas’s thought on, this is not going to be a social inclusion or women’s empowerment work, but it’s going to be a socio-economic, it’s going to be an infrastructure work as well, and it’s time that we recognize that. And this could mean paying stipends to local women, managing their Wi-Fi and maintenance and doing all the operational ground level operations, recognizing the digital caretaking that involves, that entails as a form of labor, and also not leaving out individual champions when we look at the community networks, recognizing who is anchoring these and who is pulling this together. So it’s not just going to be what they are doing well or just if the access is there, but also go beyond it and see as an infrastructure work. So these are three recommendations that I would like to add.

Rispa Arose: Thank you so much. Lilian, anything to add?

Lillian Chamorro: Well, thank you, Jess. I have a few options that we think could improve women’s participation, and one of them is establish financing funds, both for technology, but also for training with special conditions for women’s participation. For example, in closing the gaps in the basic use of technological devices and technological literacy, and facilitate access to devices. Also, adapting methods. We are a team of epidemiologists that transcend the technical approach for the community networks and also consider other women’s interests and needs related to the care and the support of the community. This diversification of roles in technical, administrative and communicative areas. The communities should promote the participation of women in technical areas such as installation and support. Also men should be involved in the administrative and financial aspects of sustainability. Also promote the participation of women in leadership roles in the communities without neglecting the recognition of the importance of the care work. I think it’s important to recognize the importance because sometimes the women, for example in our case, when we were working on establishment of community networks, of antennas, of that kind of things, many women are caring of us, are working on the food, working on the care of the people. This work is not recognized and we need to recognize that. Promote the participation of women in discussion spaces. Make women in the field of technologies visible so that they can serve as reference for other adult and young women in traditional masculinized scenarios. Reflect on care and how to approach it, recognizing its value and establishing mechanisms for its redistribution. People who require some care should be assumed as a collective responsibility of the communities and if it is necessary, resources should be allocated within the spending structures for the care of these people. Create women’s circles. We think women’s circles are a very good methodology. where women can express their concerns, opportunities, difficulties, not only about technology, but also about their participation in the life of the community. In any case, the action should be respectful of the environment and the values of the community, since we often find ourselves in communities with very different visions of gender roles, and we are not looking to break with the local issue, but rather to seek spaces for dialogue and reflection on the opportunities and inequalities faced by the women and other populations. Finally, conversations on gender must be strengthened in the entities and organizations that accompany these processes, both in men and women. If these reflections have not been integrated in our own work, it would be difficult to integrate them in the processes of accompanying the communities and in the spaces that are generated from there.

Rispa Arose: Thank you so much Liliane, and what I like about the recommendations that you have brought in is that it’s coming from working at the community level, and you know what the challenges are, and what best practice can be adopted by community-centered connectivity strategies at the grassroots. And that really fits in also very well with what Mathangi has just presented, coming from a research report that she conducted in different countries within Africa, Latin America, and Asia, and together these are really strong actions that can be taken to support the emerging and sustainability of women-led connectivity models. Now I want to go back to you Akash.

Waqas Hassan: This was a public perception survey, there was an IVR survey from around 100,000 respondents using the help of the mobile operators. So when all of this that was put into context and then this strategy was being developed, it shaped up into the form of a strategy where there is at the top level, there is this high-level steering committee, which consists of the institutional heads and led by the IT minister. And under that steering committee are six working groups, groups like access, research and data, affordability, safety and security, digital literacy inclusion. And in between these groups, each of the group has its own chair, by the way, is the implementation body, which is PTA and now GDIP has joined PTA as the joint coordinators for the strategy implementation. So with this whole framework that you can see, you can see that there was this whole of government approach that was adopted, but for the implementation itself, it is now essentially a whole of society approach that has been adopted because all of the working groups have representation, not only from the public sector, from the private sector and also from civil society, from telecom operators, from national statistical organizations, you name it and you’ll find probably find a body in one of the working groups contributing to one of the working groups. And one unique feature about this strategy is the support it extends to community networks. The working group on access, it is mentioned in the strategy as one of the DORs of the access working group that it will adopt policies around community networks. And it also mandates the establishment of at least 15 community networks across Pakistan. So this is one of the unique features of the policy that I really like and very relevant to the discussion that we are having today. Now, in the. The latest report that has just been released by GSMA on mobile gender gap, Pakistan’s gender gap has actually reduced from 38% down to 25%. And this is the largest reduction in any of the surveyed countries, which are essentially the developing countries that GSMA has surveyed. So I think in terms of lessons for any policymakers out there, I think we would recommend that first of all, please follow the inclusive policymaking process. Make sure that all of the stakeholders are on board if you want to develop this kind of a strategy, and also take the best use of other use cases, like best practice examples. And a lot of organizations are also working in this space, including ourselves, including APC, GSMA, and others. So I think this is an example from a developing country in the South Asian region, a region having the widest gender gap in the world, along with sub-Saharan Africa, of trying to make a difference in the state of affairs and actually achieving it. Thank you.

Rispa Arose: Thank you so much, Waqas, for those great points. And really, it’s good to know that this can be done, and it has been done from Pakistan. And also thank you for sharing some of the things that we can be in consideration of when thinking about a dedicated gender strategy for digital inclusion. Now I’d want to go back to Dr. Emma and Ms. Ivy, who are online as well. If you could propose one policy or practice shift to improve gender integration in digital connectivity, what would that be, and why?

Dr. Emma Otieno: Okay, thank you very much, Rizpa. And I really thank Waqas, Madhanji and Lillian for the insights that they’ve provided, which really link very well to what I’ll provide as an answer. And I like the Pakistan example, and the way they have actually gone ahead. So if I would just make it short and say, which is this one thing that we can actually do, be it at policy or as an initiative to make this improve or work better. I will go back to this, the statement that Madhanji gave during the first session of the presentation and cited Kenya as even having a national gender policy, which is true. But we have the policy and the policy has been there for quite a number of years. And it also speaks about issues of digital inclusivity and matters gender. But why are we like looking at the report that Waqas has talked about, we’re still at about 39% gender inclusivity gap. We are aware that Pakistan was years ago. So it is because of something, I think, which we’re not doing. And that is what’s going to be my answer to your question, Rizpa, is that we have not been intentional in terms of putting in these clauses in the policy regulation strategies, we are putting them in as aspects of satisfying the social or compliance aspects, but not meaningfully and intentionally putting them. And after we have also inserted and we’ve ticked the box of, we have a policy, or we have a framework, we have a guideline. What has lacked completely is measurement, the intentional and meaningful tracking, measurement, so that we can see what are the outputs out of these policies, or this policy, a clause that we have put in to promote digital inclusivity. What is the outcome, what is the impact and then how can we sustain or how can we improve and which are the areas that are paining. So when we ask anybody even in the area of policy or in the area of regulation or in any other space, we don’t have that information and Ivy said it very well, lack of even having disaggregated data to the extent that we know exactly at the point to understand what is the pain point and how do we move from here to the next step. So for me in terms of just having frameworks, most African countries are doing well because we are really complying with issues of SDG 5 of equality in terms of having those policies, but meaningfully measuring them and looking at them as contributors to socio-economic development is what has been lacking. So to summarize that answer is that we must actually start to implement things like the gender intentional digital infrastructure designs in terms of intentionally measuring, tracking, monitoring impact and feeding back through so that we sustainably improve to ensure that we are going back to closing all the gaps and of course our first answer leaving no one behind. There’s also an aspect of how can this be supported. Somebody mentioned that as I think the gentleman from the audience. PPPs bring together partnerships and collaborations because we cannot achieve this as government alone or as regulators alone or as a private sector alone. We must bring on board the financials. We must bring on board the people who are volunteering to say that they can actually translate it to their local languages. We must bring along the people who can cause devices to be cheap, connectivity to be cheap, you know, access issues. So those partnerships become the base of ensuring that actually we are making the progress in this regard. So two things, measurement and collaboration and partnership so that we achieve these aspects. Thank you very much.

Rispa Arose: Thank you so much Dr Emma. Very, very important to consider. Well said on the two critical aspects that we could be incognizant of while thinking about genomic streaming. I’d like to now hand it over to Ivy, Ms. Ivy.

Ivy Tuffuor Hoetu: Yes, thank you. And picking up from where Dr. landed, it’s true we need the metrics and we need the data. And this point was also built from the blind spots I mentioned earlier. The first thing starts from here, from the IGF here. We need from here the communique to move or go with cross-sectoral collaboration. For the past five years, multi-stakeholder collaboration has been pushed. We are doing well and I believe we can do better to address these gender-specific needs. So, different sector regulators often operate in silos with limited coordination with ministries responsible for gender, education, health, social development, et cetera. This fragmentation prevents holistic approach to digital inclusion. Our focus now should shift from one specific ministry to handle digital connectivity at least. We have now known a lot of sectors, governments know that ICT has become an enabler to all sectors. So, we cannot sit as ICT ministry or policymakers and make ICT decisions without these sectors. So, we need all of them. What IGF has done well in recent past is to involve members of parliament who are key legislators and lawmakers. I believe as they were involved, they have come in to understand the specific issues of what IGF. and other leaders, and also the IGF address. And gradually, it’s making an impact, especially in my country, Ghana. In the next decade, we need to create this platform, expand it, extend invitations to educational ministries, finance ministries, task experts. To also come to the IGF, come to the table, we need to involve them. Closely involve them. Once they sit on the table with us, understand the issues. When they are formulating their sector-specific policies, they will maybe have budgets or targets for contributing to ICT. What we don’t usually have is an infrastructure fund. Usually, this fund is taxed or contributed by mobile network operators, the ICT sector. But ICT goes beyond just the communications ministry or the regulators. It’s affecting every sector. So if a fund is created, the contribution shouldn’t just be on the limited field, the operators. The contribution should come from each of these sectors who are benefiting from ICTs. And this can help to promote community networks establishment in underserved communities. Lastly, one other thing we can do in formulating ICT policies is that stakeholder engagement. Dr. mentioned that the ICT frameworks, they are there. They’ve been developed for years. It’s the lack of implementation. Yes, they may have been prepared or drafted without extensive consultation, but it’s not too late. As we leave from here, we should involve all these stakeholders, especially those that these women… focus activists, gender activists, consumer advocates, when we are developing or when we plan to implement them, we need to involve them because they represent this specific focus group and they understand the issues. So when we involve them and we understand their basic needs, that is when we can formulate these policies to target and address their needs specifically, not making it generally, but making it specifically or addressing their specific, sorry, their specific needs. And when we do that, we get to know their issues that are bothering them, that is a concern. Then it will enable policy formulators to implement or draft and develop the right policies that will address their specific gender needs. For instance, if it’s children, ensuring their protection online. We cannot make a general, broad ICT policy just that is going to address everything equally. It cannot. We are moving from equality to equitable distribution. We need to understand the specific needs and then target them specifically. So with online, with women, we have those who are educated, those who are not educated, those who work in various sectors, and we will prepare these regulations or policies to target and address these specific needs. And when we do that, it will encourage them to use the technologies, because now we’ve gone beyond access. Availability is there. There’s availability, it’s access that we are struggling to, someone mentioned meaningful connection and access, which is key. So To be able to address this and make it meaningful, we need to understand the issues. So cross-sectoral collaboration in terms of preparing ICT policy and regulation is key in addressing gender mainstreaming issues in digital connection. Thank you.

Rispa Arose: Very, very well said, Ivy. Thank you so much for those interventions. I wouldn’t want to dilute it further. We’ll just take this time to see if there’s any question on the floor. And online, we have less than five minutes for any question, any comments that you’d want to contribute to this conversation.

Josephine Miliza: There’s no question online. Also, just a comment that women are the backbone of every society. There are fewer women taking part in the digital connectivity strategy. All they need is motivation, guidance, and support to get a lot of them in the space to make better policy. And this is from the IGF Ghana Hub. Thank you.

Rispa Arose: Thank you, thank you so much. I would really want to extend gratitude to our panelists and speakers for your honesty, your clarity, as well as your valuable contribution to this discussion. I would say that we cannot tire to talk about inclusivity because it’s towards the goal of connecting the unconnected as well as bridging the digital gap.