Day 0 Event #59 The 1st international treaty on AI and Human Rights

Summary

This discussion focused on the first international treaty on AI, human rights, democracy, and the rule of law, recently concluded by the Council of Europe. Participants from various countries shared their perspectives on the treaty’s significance and implementation. The convention aims to establish a baseline for AI governance while balancing innovation with safeguards for human rights and democratic values.

Key points emphasized included the importance of international cooperation in addressing AI’s cross-border nature and the need for a shared understanding of AI risks and opportunities. Participants highlighted the treaty’s role in setting clear principles and parameters for AI development and use, while allowing flexibility for different national approaches. The convention was praised for its focus on human rights in the context of emerging technologies.

Several speakers noted the importance of involving multiple stakeholders, including the technical community and civil society, in the ongoing development and implementation of AI governance frameworks. The treaty’s provisions for safe innovation and research were highlighted as crucial elements. Participants also discussed the convention’s potential impact on fighting disinformation and protecting democratic processes.

The discussion underscored the global nature of the treaty, with speakers emphasizing its openness to countries beyond Europe and the potential for broader international cooperation. The convention was presented as a starting point for ongoing dialogue and adaptation to the rapidly evolving field of AI. Overall, the participants viewed the treaty as a significant step towards creating a global framework for responsible AI development and use.

Keypoints

Major discussion points:

– The importance of the AI treaty as the first legally binding international instrument on AI governance

– Balancing innovation and safeguards for human rights, democracy and rule of law

– The need for international cooperation and a shared baseline for AI governance

– Implementation challenges and plans at the national level

– The treaty’s potential impact on researchers, developers and the technical community

The overall purpose of the discussion was to explore different countries’ perspectives on the recently concluded Council of Europe treaty on AI, human rights, democracy and rule of law. Panelists shared insights on their country’s approach to implementing the treaty and discussed its significance for global AI governance.

The tone of the discussion was generally positive and collaborative. Panelists emphasized the historic nature of the treaty and the importance of international cooperation. There was a sense of shared purpose in addressing AI challenges while fostering innovation. The tone remained constructive throughout, with panelists building on each other’s points and acknowledging the complexity of the issues involved.

Speakers

– Thomas SCHNEIDER: Chair/moderator of the discussion

– Işil Selen DENEMEÇ – Head of Legal Department, Digital Transformation Office of the Presidency of the Republic of Türkiye

– Ambassador Brendan DOWLING – Ambassador for Cyber Affairs and Critical Technology (Australia)

– Mauricio GIBSON – Head of Multilaterals at the International AI Policy Directorate at the UK’s Department for Science, Innovation and Technology (DSIT)

– Allison PETERS: Deputy Assistant Secretary in the Bureau of Democracy Human Rights at Labor, U.S. State Department

Additional speakers:

– Nigel CASSIMIRE, Deputy Secretary-General, Caribbean Telecommunications Union

– Lina VILTRAKIENE, Lithuanian ambassador for digital economy

Expanded Summary: Council of Europe AI Treaty Discussion

Introduction

This discussion focused on the recently concluded Council of Europe treaty on AI, human rights, democracy, and the rule of law. Participants from various countries shared their perspectives on the treaty’s significance and implementation challenges. The convention aims to establish a baseline for AI governance while balancing innovation with safeguards for human rights and democratic values.

Key Themes and Discussion Points

1. Significance of the AI Treaty

The discussion emphasized the importance of the treaty in addressing AI governance. Ambassador Dowling from Australia succinctly captured the core tension addressed by the treaty: “We all see the upsides of AI. We all see the benefits for development, for economic opportunity. But with every new phase of digital technology, we’ve seen human rights, the rights of women and girls, the rights of freedom of speech, democracy jeopardised.” This comment set the tone for much of the subsequent discussion.

2. Balancing Innovation and Safeguards

A central theme was the treaty’s aim to balance AI innovation with protection of human rights and democratic values. Allison Peters from the U.S. State Department emphasized this point, stating, “We don’t want to do anything as it relates to regulation that cracks down on innovation, doesn’t allow that innovation to happen, and so certainly, with this convention, we see a convention that allows us to harmonize various different international approaches to AI, and one that still allows companies to innovate.”

Mr. Gibson from the UK government provided insight into a specific provision addressing this balance, noting an exemption for research and development: “In the convention, there is a sort of, I think I touched upon it in my statement as well, but an exemption for research and development. So what that doesn’t mean is it just excludes all and every single form of research and development from the convention, but it creates a sort of a limitation and a safeguard.”

3. International Participation and Perspectives

David Fairchild from Canada highlighted the effort to involve a diverse range of countries in the treaty process: “It’s not the UN, but at the same time we felt it was extremely important to try to get as many member states from as many regions involved in the process.”

Isil Denemec from Turkey introduced the perspective of how a country’s specific geopolitical context shapes its approach to AI governance, stating, “Turkey’s unique position, geopolitical and geostrategic position influences its approach to global AI governance.”

Tetsushi Hirano presented Japan’s perspective, emphasizing their commitment to human-centric AI and the importance of international cooperation in AI governance.

4. Implementation Approaches and Challenges

Countries are taking varied approaches to implementing the treaty and aligning it with national AI regulations. Mr. Gibson explained the UK’s approach, which includes the AI Safety Institute and efforts to align with international frameworks while maintaining a pro-innovation stance.

Mr. Thomas Schneider introduced the Huderia (human rights democracy and rule of law risk and impact assessment methodology) as a tool for assessing AI systems’ impacts. He also clarified the distinction between the Council of Europe and the European Union, emphasizing the Council’s focus on human rights, democracy, and the rule of law.

5. Addressing AI-Specific Challenges

Lina Viltrakiene, the Lithuanian ambassador for digital economy, highlighted the importance of combating AI-driven disinformation and protecting democratic processes. She also mentioned the Vilnius Convention and Lithuania’s role in the treaty signing.

6. Future Developments and Adaptations

The discussion touched on the potential for future adjustments as new countries join the treaty. Nigel Cassimire from the Caribbean Telecommunications Union inquired about the process for additional countries to join in the future.

Additional Points of Interest

1. A question from the scientific community addressed the potential impact of the treaty on daily coding practices.

2. The brief discussion on disinformation highlighted its relevance to the treaty’s objectives.

Conclusion

The discussion underscored the global nature of the treaty and its potential for broader international cooperation. Participants viewed the convention as a significant step towards creating a framework for responsible AI development and use, while acknowledging the need for ongoing dialogue and adaptation in response to the rapidly evolving field of AI.

Mr Thomas SCHNEIDER: So, can you hear me okay? Many people can hear me? Okay. All right, let’s start. So, as you know, this early morning session is about the first international treaty on AI, human rights, democracy, and the rule of law, which has been concluded earlier this year. And of course, the aim is to cope, to help us cope with the risks and opportunities that AI offer or bring. And I will not go into detail about risks and opportunities because we all by now more or less know what these are, at least for the time being. So, let’s start right away with Ambassador Dowling from Australia. Australia was also actively participating in the negotiations of the treaty until this March. So, how does Australia view the balance between innovation and safeguards so that we have a robust and resilient innovation thanks to AI, but also make sure that human rights, democracy, and rule of law is protected? Thank you. Look, that’s the key question I think we’re all grappling with. We all see the upsides of AI. We all see the benefits for development, for economic opportunity. But with every new phase of digital technology, we’ve seen human rights, the rights of women and girls, the rights of freedom of speech, democracy jeopardised. So, I’m really positive about the Council of Europe’s effort here to get ahead of the curve, to say that we will step in early in the development of this technology and its far more ubiquitous use and try to set some guardrails, protections for rights, speech, the preservation of democracy. I think we made the mistake in earlier phases of technology when we saw software development in the 90s, when we saw social media in the early 2000s, in saying we trust the technology world, we trust commercial entities to prioritise safety and rights online. learnt the hard way that that’s not the case, that we shouldn’t expect commercial entities to be the guardians of rights and privacy, that that’s a role for governments and civil society to work together on. AI is a fast evolving area of technology but I think what this treaty does is set some very clear principles and parameters that say here are the expectations that governments have on the development of this technology, here is how we expect all platforms to preserve rights, privacy, ensure their technology is not misused and so I think acting that early in a way that’s not overly onerous or prescriptive, I think the treaty strikes at right balance where it will not stifle innovation. We feel that it’s a very commendable effort to try and set those parameters early in this particular phase of AI to ensure that we’re not looking back in 10 years time saying well we wish that we’d done more in the early phase to preserve rights and democracy. Thank you very much Ambassador Dowling. Now we go to to the West, to the United States. We have Alison Peters here, Deputy Assistant Secretary in the Bureau of Democracy Human Rights at Labor, the State Department. So Miss Peters, the US was very active in the negotiations on that treaty as we have been able to read also in some journals. How do you see the treaty promoting respect for human rights democracy around the world, not just in Europe or in the United States and how does the convention, the way it is in front of us now, reflect the United States approach for rights respecting AI?

AUDIENCE: Well first and foremost just a huge thank you to you Thomas and your incredible team. I think it has been several years of negotiations and close partnership I think across governments engaged in the Council of Europe and it is because of your leadership that we’re really here today talking about the fact that many of our countries have now signed the world’s first ever treaty on artificial intelligence and that’s quite a feat. I think second we really need to acknowledge that in a context in which we continue to have debates in the United Nations and in the broader multilateral system in terms of the applicability of international law, particularly international human rights law, it’s quite striking that the world’s first ever treaty on artificial intelligence is focused on human rights, democracy and the role of law. For the United States you know we engage actively in this negotiation process because we see a couple of key benefits in having a legally binding convention on our artificial intelligence. And of course, Ambassador Dowling laid out some of the both opportunities and risks that we see as it relates to artificial intelligence and emerging technologies. But we know first and foremost, having a convention will really help us set a shared baseline on rights respecting, use, development, design of artificial intelligence. And that shared baseline goes beyond just the United States and other countries sort of in the global north, right? That shared baseline is a shared baseline that is really global in context, that is applicable to every single country that is here at IGF and beyond. But we know, given the debates that we’re seeing in terms of human rights and their applicability to emerging technologies, that having a sort of shared baseline amongst democracies on rights respecting design, development, and use of AI is critical. Certainly, the benefits don’t end just with implementation of this convention, as we’ve seen, for example, in the Council of Europe context where we have the Budapest Convention on cybercrime, having an instrument in which our governments can focus on shared cooperation, having a mechanism for shared cooperation, is incredibly useful, is incredibly salient as we continue to debate issues around cybercrime in the UN system now. The follow-up cooperation mechanisms that we expect with this AI convention will also, I think, provide a really important opportunity to advance our shared efforts and promote human rights, democracy, and the role of law on the context of artificial intelligence, including allowing us to really share best practices across the board, across governments, across regions. Third, I will say, you know, you asked the question on how this helps advance sort of the US priorities as it relates to artificial intelligence, and as we look at the need to place safeguards on AI, on AI systems, and the companies that develop them, it’s also really critical that we preserve the space for innovation, right? We don’t want to do anything as it relates to regulation that cracks down on innovation, doesn’t allow that innovation to happen, and so certainly, with this convention, we see a convention that allows us to harmonize various different international approaches to AI, and one that still allows companies to innovate, to be creative, to create new AI systems that help really advance the opportunities as it relates to AI, but also help crack down on some of those risks that we’re talking about. In our system, all of these priorities are bipartisan in nature, so even though our government is changing over quite soon, this is a priority that we see across the board, across political parties in our countries, and we look forward to working with the many governments on this stage, but hopefully other governments here at IGF who will join us in this process. So again, just a huge thanks for having us today.

Mr Thomas SCHNEIDER: Thank you. And I think at one point that is important is that it’s not just the convention and the parties, it’s the bigger cooperation setting around, which is something that we have been able to see with the Cybercrime Convention, with the Budapest Convention, as an important factor as well. Let us now move to another ICE country. Canada, Mr. Fairchild, you also participated very actively in the work on the treaty, helping us to count the hours that we have left to find compromise on everything. So if you could share your insights into how international cooperation can strengthen the treaty and ensure that it reflects the shared values of democratic nations.

AUDIENCE: Thank you very much, Ambassador, and good morning to everybody. It’s great to be here, day zero, it’s always a bit of a cold start. So thank you, and I think first of all, I’d start by thanking the Council of Europe, we have some of the staff here from the actual process in the room. It was a very long journey, but I think at the end of the day, the result is a very positive one. speaks for itself. So I think, and just to reflect a bit of the earlier comments, I mean this is the first international treaty on an emerging technology. I think it’s really important to reinforce that, you know, it’s the Council of Europe that brought forward this idea and has seen the creation of effectively what is the first legally binding instrument that is creating obligations on signatories to uphold certain values. Human rights, respect for democracy, and the rule of law. I think that’s an important statement in and of itself because it’s very different from a lot of the other work that’s going on in other multilateral fora which are speaking more on the normative basis. And I think which is why Canada, who’s not a member of the Council of Europe, it was an observer to the process, and others on the floor here, really sought to participate because this was our opportunity to frame at the international level effectively the first legally binding instrument. So from our perspective it was crucial to be there as an active participant on the delegation. We spent a lot of time trying to involve as many states as possible and so I think this is another important element to bring to the to the debate. It’s not the UN, but at the same time we felt it was extremely important to try to get as many member states from as many regions involved in the process. And so throughout the nearly two years of negotiation we saw a gradual increase not only in the number of participants but in fact signatories, hopefully at the end of the day. And so a cluster of observer groups in and of themselves. Could you bring me also an earphone for me?

Mr Thomas SCHNEIDER: Okay. And I think it’s an important statement. And we continue in that regard. So even past the end of the negotiations we’re seeing more states actively coming forward, trying to become part of the process. Of course post negotiations it’s sort of… have to take the treaty as it is but we are seeing increased interest particularly the Americas showed a high level of interest Asia as well we had Japan who was one of the observers but more countries at this point seemingly coming forward with an express interest and why is because they also have at the national levels everybody is come fascinated with the question what to do about AI. AI is a step change in the environment it’s an enabling technology which means if you haven’t regulated legislated or created frameworks to manage AI it is going to impact pretty much every sector of your government and every public policy area and so for us from the beginning we were also part of the previous process that the Council of Europe ran for the two years prior to the launch of negotiations this was all about framing and it’s important member states have obligations and so that the context of this framework instrument was to create a baseline and this is an interesting second point is this is one of the few areas where we were negotiating international treaty where there was frankly no floor most member states don’t have AI legislation so it wasn’t about leveling up at the international level it was actually about creating the baseline level to which member states have to now meet and so was it a very unique opportunity for Canada who is quite advanced in its own legislative and regulatory frameworks to help in fact you know create that floor and so I think we were quite successful always negotiations is about concessions but the end of the day I think you know could we have had an in a stronger instrument yes could we have had a weaker instrument yes and I think at least from Canada’s perspective there was a fine balance to the negotiations but at the end of the day what we have successfully created our baseline obligations that member states now have to take away and as they develop their own regulatory and national legislative frameworks they must meet these obligations I think that was a critical reason for us to be a part of this thank you thank you very much

AUDIENCE: David, I’m now nicely cabled and hearing myself several times with several loops. So yes, I think this is an important element that you named, that we didn’t start like comparing our national bases, but we had to basically set this floor, set this baseline from scratch, which was a challenge, but it was also an opportunity to seize. So now we go to a European country for once. We have Izil Deminec from Turkey, Turkey. You were also actively participating in the negotiations. How is Turkey planning to implement this convention? How are you going to organize yourself so that you address the challenges posed by AI and help your country to stay innovative at the same time? Thank you, Thomas. Thank you for the organization and the invitation as well. I hope my voice is… So as a member of the Council of Europe, Turkey recognizes the transformative potential of AI and obviously it brings with it a profound batch of legal and ethical questions. And the rapid spread of AI made it essential for everyone to take proactive steps to uphold AI principles at every stage of the AI lifecycle, which we were addressing at the CAI as well. To establish a robust framework that fosters ethical, sustainable, and innovative AI while addressing the unique challenges that AI poses, we have adopted the National AI Strategy back in 2021. It was drafted together with the Digital Transformation Office and the Ministry of Industry and Technology, and it is built upon six strategic priorities, which includes regulating to accelerate socio-economic adaptation. one of the ways Turkey is planning to address these challenges is through strengthening the legal framework. This is being done through amendment of existing laws or adopting of new ones as necessary and alongside the AI rules and regulations we are also working on our frameworks concerning both personal and non-personal data because we cannot separate the two from together as data is an integral element of AI technologies. So as per the strategy Turkey’s core objective focus on establishing an agile and inclusive process at developing governance mechanisms that promote impartiality, privacy, ethical oversight and algorithmic accountability while we also enhance data capacity to assess AI socio-economic impacts. The strategy also underscores the importance of fostering innovation which was one of our concerns at the CAI as well and ensuring alignment with international ethical standards. So I will be very briefly listing some of the key measures to support these goals. Aligning national AI regulations with international frameworks and commitments to maintain consistency with global governance structures. Creating an AI impact assessment framework which we are currently dealing with at the CAI to evaluate ethical social and economic impacts of AI systems. Preparing and implementing guidelines for algorithmic accountability and explainability to ensure transparency in automated decision-making and encouraging capacity building initiatives and public awareness in AI use. These are some of the issues that we are actually dealing with at the CAI. So it is essential to emphasize that legal challenges, these legal challenges require legal solutions beyond just legal considerations because we need to encompass social, financial, political and other dimensions at the same time. And Turkey’s unique position, geopolitical and geostrategic position influences its approach to global AI governance. That’s why we have been actively monitoring and participating in international efforts. shape AI governance frameworks, not just at the CAI, which is also one of the strategy’s six priorities. This commitment is also clear in our active involvement at the CAI processes from the very beginning, since the CAHI, which led to the development of the first global legally binding AI treaty. So we are aware that we are at the forefront of disruptive technological change and it is evident that AI holds immense power, potential to shape our societies for the better. But how can we do this? We need robust ethical and legal frameworks that guide the process. So the framework convention emerging from the CAI process represents a pivotal step in this direction, which embodies the collective efforts of nations to ensure AI aligns with human rights, democratic values and shared global practices. We must recognize that AI knows no borders and its impact is inherently global, therefore a legally binding international instrument is not just desirable, but also essential to address these issues, to create a unified approach. Well, this instrument we hope will ensure accountability, transparency and fairness, while fostering trust among stakeholders to also help the development of this technology. Therefore we will continue to be actively contributing and remaining in the CAI processes and other alongside our other work on AI and other technological advancements. Thank you.

Mr Thomas SCHNEIDER: Thank you very much Isil and I think it’s also important to raise the importance of data and data governance. The convention of course deals mainly with AI, but you find hints to data across the convention and if you in particular if you look at the explanatory report, then of course you find references to appropriate data quality and other issues related to data. Now we have an online participant. I hope the connection works. It is a colleague Tetsushi Hirano from Japan. Japan has been also a very active participant in the negotiations on that treaty at the Council of Europe in Strasbourg and but Japan has not only been a very active at the Council of Europe, Japan has also led other important initiatives like the G7 Hiroshima framework and of course Japan is also active in other processes like OECD and others. So Tetsushi, I hope you are there. My question to you would be How does Japan’s approach to AI governance align with the principles being proposed in this treaty? See whether the connection works … Any information from the… it’s coming? What’s the time in Japan now? What is the time difference? It’s already in the afternoon. Yeah, or we can if there’s some difficulties we can we have one more physically present human being here supposed to talk. We can switch the order, Vadim. No, he is coming. Okay. So let’s… Yeah, now we hear you. Hello. Did you hear my question or should I repeat it?

AUDIENCE: Okay. Okay, excellent, so we’re happy to listen to you now, Tetsushi. If you can speak up a little louder, your voice is a little low, so that may help. Do you hear me? Yeah, okay. Okay. So I’m very happy to be able to participate in this session to discuss this important international treaty with familiar faces from the negotiation. And I would like once again to thank Thomas and the CHI secretariat for their outstanding work and support throughout the negotiation process and to congratulate the states and the EU for signing the convention. In answer to Thomas, I think it is symbolic that this treaty was adopted in an institution that was founded on the bitter lessons of the Second World War. At the heart of what the Council of Europe stands for are common values that Japan shares. This is one of the reasons why, after the fall of the Berlin Wall, Japan, together with Canada, the United States of America and Mexico, became an observer in the Council of Europe. These common values are crystallized in different ways in international law and in the constitution of the 57 states that have agreed on the frame of conventional artificial intelligence. The aim of the convention is to ensure that the activities within the life cycle of artificial intelligence systems comply with existing international and national legal obligations. This means that Japan, along with 56 other countries, has agreed on a framework to marry the shared values into the future where AI systems will be used in every corner of society. Some of the principles of the framework are transparency, equality, privacy, reliability, which are developed taking into account the socio-technical characteristics of AI technology. These principles will be operationalized in the risk and impact management framework. In my view, one of the challenges for the implementation of this treaty is the operationalization of these principles in accordance with existing legal obligations. It seems that the key lies in building capacity on the ethical, legal, and social implications of technological characteristics of AI, such as explainability or accuracy. This summer, Japan launched a new discussion on the future domestic framework for AI regulation under the AI Strategy Council, taking into account the new rules, guidelines, as well as recent achievements in the international forum, including the Hiroshima AI process. At the same time, we are accelerating the internal process of signing the convention. I am convinced that it is to possess the right balance of innovation and regulation. Thank you. Wait a second, the connection has turned really bad. Sound and zoom is okay. So you can tell us what, yeah, okay. We’re having a little bit of a technical issue here. If you could repeat the last few sentences that you said because we have missed it, unfortunately.

Mr Thomas SCHNEIDER: Thank you. Well, we can’t hear you. We don’t hear anything. It’s just very faint, faint noise, but our 500 people technical team is already working on it. So. Um. I have to wait. Okay. Okay, thank you, Tetsushi. I think we’ve heard some elements from you, at least. There seems to be a problem that here in the room, we can’t hear you anymore. Maybe let’s use the time and we’re happy to take you back in the discussion. Let’s move on to Mr. Gibson. You may not have anything to do with the famous guitar brand, but you work for the UK government. So, yeah. In the UK, now that you signed the convention, how do you use the convention to also move things ahead on national level? And how are you seeing international cooperation thanks to the convention, but also in other frameworks? What are the UK’s plans for the coming months?

AUDIENCE: Yes, thank you. Thank you, Ambassador. Can people hear me? go ahead yeah people hear me yeah now we can really thank you very much thank you ambassador yes not not quite linked to the Gibson guitars I do get that and I do get Mel Gibson as my uncle but yeah I am yes and as to your question I think a good way of framing this is looking at starting off by considering the UK’s approach to air governance obviously we’ve had a government recently and we’re thinking about the opportunities of AI and really want to capitalize those as well in order to turbocharge economic growth and you know without a sort of sense of an understanding of the risks mitigating risks tackling key risks as well we won’t be in a position to fully capitalize or you know capture the full potential of the technology as well and so that’s the kind of key framing I think is really important when we go into discussing the Council Europe AI treaty as well as other international governance frameworks really understanding that working together cooperation and setting global baselines due to the transcendental cross-border nature of the technology is fundamental and really harnessing the potential of every sort of input from different countries civil society and the private sector as well whilst recognizing that in order to really capture the the benefits opportunity and harness innovation you need a technology that is trustworthy that those risks are understood that you can have that safe secure trustworthy AI and really capture that as well so I think that’s been a sort of key framing I think for us as well at the international level we’ve really not been able to highlight the it’s such an extent though but when the UK we sort of had the first AI safety summit in Bletchley that paved the way for a great deal of discussion on a lot of these risks and what we need to do to mitigate that. And in doing so, we were very keen to really reinforce in the Bletchley Declaration the importance of safe, secure, trustworthy AI and building that understanding of risk as well. With that in mind, we were also very keen to really harness the possibilities of securing innovation for that. And so bringing Frontier AI Labs and companies on board in that process has been really fundamental and really highlighting the importance of getting voluntary commitments, pushing the conversation along with a range of sectors to really ensure that the public and different sectors can buy into that technology. That has continued in other areas of international AI governance as well. We’ve really reflected this in the UN environment. So the Global Digital Compact, we really wanted to reinforce the importance of a multi-stakeholder input. In the UN resolutions, which were agreed in 2024, we wanted to really push that there were flexible and differentiated approaches. And that’s kind of reflective of the UK approach, that in order to get that trust, you need a proportionate approach to governance, and that’s a fundamental recognition. So you get the balance between innovation and safety. Coming to the Council of Europe specifically, though, I think this has been an opportunity for us to really reinforce the UK’s value add in the conversation. So we have taken a sort of very, reflecting on the proportionate amount, a balanced, targeted, agile approach to AI governance. So we look at things sort of sectorally in our regulatory approach at the moment. So allowing different regulators to engage with the technology and regulate technology in a sort of more light-touch approach, ensuring that these all work together and really harness the potential that way. However, we are thinking about sort of building on that and recognizing the role of innovation with our new Regulatory Innovation Office, with a model for responsible innovation, supporting the technology. And this is the sort of approach that we have reflected in the Council of Europe AI Treaty negotiations as well. well to bring us to that point. So the UK was really keen to reinforce the merits of balanced language, so ensuring that there is not too much prescription in the detail, so being too over prescriptive might make it a bit too challenging for a lot of different countries to get on board. So looking at the range of different countries here, I think we recognize that we want a range of different regulatory approaches to be recognized in the Council of Europe AI Treaty, and also reinforcing the importance of new and other countries who are a bit nascent in the regulatory approach to AI to come on board as well. However, I think the importance is balancing that with clarity, and we were really keen to push the point about clarity in text. These are some legal obligations, you know, involving human rights. We want to be really clear that people understand the technology, and that comes back to my point about trust. If you don’t trust, we can’t really capture the technology as well, and I think this was reflected in a couple of really interesting provisions in the convention as well. So a provision on safe innovation, another provision which ensured that research and development could be safeguarded, and wasn’t necessarily always fully in line, and making sure that that can be continued despite or in spite of it at the convention as well, but really making sure that’s crystal clear. And we wanted to make really clear that we could be a bridge as well as the UK, so talking about that proportionate approach to governance, we recognized that all these different regulatory approaches would be needed, and towards the end when it might have looked a little bit challenging to get agreement, we were really keen to be pragmatic, provide a sort of basis in bringing a group of people together on the sort of final throws in the 11th hour of the negotiations, to come together and find those areas of overlap on the more challenging areas, so how we could apply the convention and make sure we could get all the relevant countries on board so that this could be a successful global treaty. So we were really keen to really be pragmatic, to really reinforce those points about trust as well. well, and I think that’s reflective in our international AI governance approaches.

Mr Thomas SCHNEIDER: Thank you very much Mauricio, and indeed it was not easy, and hopefully we get a discussion, a few minutes for discussion also in the end. The idea to make something that holds across time with principles, but also to have it sufficiently clear, but not too over-perspective in detail, was and is not an easy task and will also be a challenge in implementing it. Now I wonder, we would have another very active participant, but apparently, Argentina is a country that was also very active, but apparently the connection somehow didn’t make it across the ocean, but just to say that Argentina would have been an example of one of the Latin American countries that have been participating, there were about 5 or 6 Latin American countries that were already doing the negotiation, active in the work on this treaty. But we’ll see whether we can connect to her later. So with this, we have gone through the panelists, unless I missed somebody, which doesn’t seem to be the case. So we could actually use a few minutes to also allow the people in the room and also online to make comments and questions. We have a mic here, so just stand up and make yourself heard if you want, and we’ll try to have some interaction also with the audience here. Yes, please present yourself and then continue. Well, thank you very much. My name is Dr. Rafael de Lis Aguirre, I’m at the Max Tilburg Center in Germany, and as a member of the scientific community and a developer, what impact do you foresee that such a treaty will have for us and our daily lives coding? Thank you. Thank you very much. This is a very good question. Any of you who wants to reply to this? Allison?

Allison PETERS: Well, thank you, first and foremost, for the question. I think I can only speak for my government, but I also know I speak for several governments here, which is that throughout the negotiation process, with big recognition of the fact that this is a legally binding convention, making sure that the technical community could actually participate and that we had on our delegation technical experts that were leading the negotiations was mission critical. I think for us, and I think my colleague from the UK spoke about this as well, making sure that we have a convention that is technically rigorous, that is practical, that is clear in terms of how it can be used, not just by governments, but also the technical community was really a big priority. I won’t talk about every single provision of the convention, but I think for you and colleagues in this space, you know, one of the key things that that we’re working on now is actually building out a risk assessment framework. And again, we talked, I think every one of the panelists talked about the fact that this convention would be really a sort of shared baseline in systems that may not have an AI risk assessment framework. So in the United States, we have our NIST Risk Management Framework, or RMF, as it’s known by the technical community, that actually helps us do assessments of AI systems and managing risks and putting then in place technically rigorous safeguards that could manage those risks. But other systems don’t have a NIST Risk Management Framework or don’t have risk management frameworks that necessarily harmonize across systems. And so we are actually working to negotiate a risk assessment framework that could be used across the board, whether that be by governments or individuals in the technical community. And if you’re not already tracking that process, we’d love to make sure that you are able to engage and that you’re able to follow the negotiations, because I think for folks like you, that’s gonna be a really important thing that we built out, known as the Huderia. And for us, again, making sure that that tool is technically rigorous, but also practical and interoperable across systems is really top priority in the current moment.

Mr Thomas SCHNEIDER: Yeah. I would just also, yeah, reinforcing everything my colleagues from the U.S. said. And I think another thing to touch upon is in the convention, there is a sort of, I think I touched upon it in my statement as well, but an exemption for research and development. So what that doesn’t mean is it just excludes all and every single form of research and development from the convention, but it creates a sort of a limitation and a safeguard. And in certain conditions, those undertaking certain R&D, which might impact human rights, can continue to do so with the necessary safeguards in place. So that was a really… hotly focused on area that we really try to invest a lot of time so we could make sure that people like in your community Have that space to continue doing things where there might be implications for human rights and Really continuing that as such so that’s a particular area which I think is important to highlight The same with the safe innovation space or creating a space for sandboxing and encouraging different regulations to engage in sandboxing which can Do things safely and in that space really creating the environment for that as well Just just to add to this one one comment and indeed one thing is to have some kind of Harmonization or at least interoperability on legal terms But then if you’re supposed to implement this and of course the convention Because it’s supposed to hold for a few years 10 hopefully 20 needs to be on a very general level And this is why this risk Implement Impact and risk assessment frameworks are so important and the second deliverable that the CHI this Council of Europe committee Is delivering is this Huderia the human rights democracy and rule of law risk and impact assessment methodology The key requirements on a very general level are part of the convention We’ve just adopted there what we call level two document Which is a 20 page guidance and non-binding a guidance document about how to do Risk and impact assessment and next year. We’ll be working on a much more detailed Document with questionnaires and so on they’re going into fairly detailed level in and this is the important thing in cooperation with the IEEE with I EEC the International Electoral Technical Committee with ISO so also to build the bridge between technical standardization institutions that Programmers are probably more familiar with to the legal standardization body, so that’s a very good question. Thank you very much. Thank you next

NIGEL CASIMIR: Hi Thank you Good day, I’m Nigel Casimir from the Caribbean Telecommunications Union, and I’m wondering two things. One is, are there any arrangements for additional countries that might want to accede to this convention? And secondly, given that, you know, it was created, I heard a lot of talk about negotiations and so on, and first in the Caribbean, you know, we would think that the countries that got this thing started would basically be kind of like-minded, so I’m wondering, apart from the processes for additional countries to accede, are there any thoughts about, as additional countries come in, to make allowances for, I don’t know, adjustments to the treaty and that sort of thing? So I guess future processes in general.

Mr Thomas SCHNEIDER: David, thank you very much, Nigel, and good to see you. Very good question. Maybe one word. The Council of Europe has got nothing to do with the European Union. The Council of Europe, as was said, was like the UN of Europe, was created after the Second World War to try and bring peace in a sustainable way to Europe through securing human rights, democracy, and rule of law. It has no economic component, and it has 46 member states now, whereas the European Union, the members of the EU are all part of members of the Council of Europe. And the Council of Europe has this unique opportunity to not just develop soft law and hard law standards for its members, but it can also include others, other countries. We’ve already had 11 in the negotiations, and any member states, any country in the world that lives up to certain standards on human rights, democracy, and rule of law can become a part of the process. We have contact with a number of countries in Africa, Asia, Latin America, that are in contact with the Council of Europe to become part of the process, also to keep working on the Huderia, but also to become a potential future party of the Convention. happy and if you turn around the tall man behind you is working for the Council of Europe, Vadim, he’s willing to get into contact with interested countries or stakeholders from other countries to actually broaden the basis and the example of the cybercrime treaty actually shows that, I don’t know the exact number, but the number of signature of parties is around 70, but the cooperation actually, one is to have a sign and ratify the convention, the other one is to cooperate around the process and there’s way more than a hundred countries that are in contact with the Council of Europe and at least take over some of the elements which helps to interoperability of legal systems and then also the implementation mechanisms across the world. So you’re very warmly invited to spread the message that this is a treaty that is open, that can be and countries can join to become part of the treaty but also to be a part of the wider process on cooperation and exchange of best practice. Yes, David?

AUDIENCE: Sorry about that. Nigel? Yes? Okay, good. Great, great to hear from you Nigel and everything he said plus one. I think the important element, at least from Canada’s perspective, is we were not a member of the Council of Europe so our participation was crucial simply because the concerns that Canadian legislative basis is different from Europe, is different from the EU, and so we had equities we obviously wanted to protect, values we wanted to bring forward in the process. I mean on a personal basis, would I have loved to have seen more observer states, more blocs, whether it be CARICOM or others, participate in the process, the negotiations? Absolutely, but we can’t change that. But what we did to make sure in the process the negotiation was to create a conference of the parties. And so this is sort of, once the treaty comes into force, which is a very low number of member-state signatories and ratifications, it’s an annual process for the members to, the members of the convention to come together and consider the treaty itself. It’s a multi-stakeholder forum, which means participation from the technical community, the civil society, private sector entities, to continue to review. Because inherently we understood that it’s very hard to create international legislation that is future-proof. And so in a sense, whether you are part of the negotiations or not, obviously down the road, you know, a member of CARICOM or CARICOM itself, I mean, being a part now and signing up allows you to be part of the ongoing dialogue. And I think the important element here is the treaty itself is very slim. It’s not very long. It’s not very detailed. It was about, you know, as the expression goes, it doesn’t matter what color the cat is as long as it catches mice. And what this treaty was trying to say is we were just trying to make sure we all understood what the cat should look like, not the color. The color is a national effort. And so, you know, Canada has very advanced regulatory regimes. It’s working on a piece of legislation. You know, other countries on this floor don’t even have legislation or, you know, frameworks are of a different nature. And so the whole point is whether you’re from Asia or from Africa or North America, as long as we adhere to basic principles, your cat can look very different, but as long as it’s still a cat. Thank you very much for this very good picture. Thank you. Next,

LINA VILTREKENE: please. Hello, my name is Lina Viltrekene and I am Lithuanian ambassador for digital economy. I would like to thank very much the Council of Europe, the chair and the panel for having this discussion on this first international treaty on artificial intelligence, human rights and democracy and rule of law. So I intentionally said the long name of the treaty. because the treaty was opened for signature in Vilnius, in my capital, and we are proudly calling this treaty as the Vilnius Convention. Proudly because it was also signed during the presidency of Lithuania to the Council of Europe, and it coincides very much with Lithuania’s priorities during that presidency, including strengthening democracies, including protecting human rights, but also including fighting disinformation. And I wanted to emphasize this important element in the context of indeed alarming rise of disinformation, which really undermines human rights, and particularly of those groups of society that are most vulnerable, like children, like seniors, like people with disabilities. So I just wanted also perhaps to raise a question, how do you see how the Convention could contribute to fighting disinformation, particularly AI-driven disinformation and harmful content online?

Mr Thomas SCHNEIDER: Thank you very much, and first of all, thank you, Lithuania, for your support, and we’re very happy throughout the process, also as a president of the Council of Europe during that period, also for organizing EuroDIG, the European IGF in Vilnius this year, so we’re very happy that this is going to be, the Convention is going to be known as the Vilnius Convention. On the question on disinformation, there’s one very general reference, which is very fundamental, but it remains on a very general level about states’ obligations to secure the functioning of democratic processes, and it doesn’t go much further. into detail, because this is very difficult. It’s also very different from from country to country how democratic processes are organized. So yeah, any short references maybe on how in your countries you deal with democracy? Maybe Isil from Turkey, do you have a…

AUDIENCE: Thank you. I just wanted to, I think you brought up a really important topic. Disinformation can definitely impact democratic processes and this is one of the things we’re trying to protect with this convention. And as our chair of CAI, previous chair of CAI has mentioned, this is a framework convention so it’s guiding the signatories to adopt measures or laws to prevent issues like this from happening. Well, as you correctly mentioned, the rise of disinformation can actually manipulate public opinion and in turn affect the democratic processes. So this can be something that can be developed by signatories in a more detailed manner in their own legal regimes, I believe. Thank you so much. I think this is one of the problems that your government, our government, we’re all facing this challenge of how AI changes the information environment. I’d first say that foreign interference, disinformation, these are not new challenges. These are long-standing issues. AI does supercharge the ability for actors to use tools. We’ve seen foreign states already using AI tools to conduct information operations in many countries. At the same time, we can overstate the risk of disinformation. We can sometimes see countries use the idea that disinformation… is prevalent to censor political speech so it’s a very careful balance to be struck in terms of saying yes there is a threat to democracy to democratic participation however there is also the risk that it is this this narrative is used to censor free speech so I think a really core obligation on us all is to build trust to build an ability for our societies to critique the information environment to be able to engage in information so that we’re not just believing stories that are out there but this is not an area that governments should dive in too deeply too quickly because I think there is a rush that it can it can be over corrected and in fact we can stifle the information environment in the desire to address this information thank you unfortunately 130 seconds for a listen because time is up we need to free the room I guess and also to the next one’s just 30 seconds I just really wanted to thank Lithuania for your leadership in allowing us to have the signing in your country and to come back to the point that Thomas started with while this is not a dedicated convention on AI exacerbating disinformation or broader information integrity around the globe it does deal with the use of AI systems by public sector actors meaning governments right and so it to the point that Ambassador Dowling made on this delicate balance between both addressing risks of artificial intelligence and safeguarding fundamental freedoms I think we strike that right balance overall in this convention and coming back to the point where we all started I mean this is a shared baseline so as we look to the UN system other multilateral systems where governments are perhaps not striking that right balance and debates around disinformation and broader information integrity having this shared understanding is going to be mission critical as we take forward those negotiations in those systems so while it’s not exclusively dealing with issues around information manipulation I think it’s certainly quite critical that we have this shared understanding to start.

Mr Thomas SCHNEIDER: start just finish with one reference to an online question about the relationship or how the Council of Europe treaty can be useful to the bigger global UN environment also helping to implement the provisions from the GDC this is something we’ll have more more sessions on GDC so you’re very happy also to to think about how this convention can contribute to implementing the GDC and as Alison has said and many others have said this is a baseline it this convention will need to be and is already complemented by other more specific soft and hard law that follows the logic of the convention so currently for instance that the Council of Europe does work on the impact of generative AI on freedom of expression on on democratic deliberations in a country this is just one example, so thank you very much for your attention. Yeah, we’ll look forward also to engaging again with countries that are not yet part of the process. You’re very happy to come to me or to Vadim or to anyone that you’ve seen here on the panel. We’re all very, very eager to connect with countries that we are not yet connected about the convention. Thank you very much and enjoy the rest of the day online or offline, thank you. Thank you. Thank you.

Agreement Points

Importance of the AI treaty as the first legally binding international agreement

Mr Thomas SCHNEIDER

Allison PETERS

First legally binding international treaty on AI

Sets a shared baseline for rights-respecting AI development

Speakers agree on the significance of the AI treaty as the first legally binding international agreement, setting a shared baseline for AI governance.

Balancing innovation and safeguards in AI development

AUDIENCE

Allison PETERS

Balances innovation and safeguards for human rights

Harmonizing various international approaches to AI

Speakers emphasize the importance of striking a balance between fostering AI innovation and protecting human rights and democratic values.

International cooperation and alignment of AI governance

Mr Thomas SCHNEIDER

Allison PETERS

AUDIENCE

Open for participation from countries worldwide

Creating mechanisms for shared cooperation and best practices

Aligning national AI regulations with international frameworks

Speakers agree on the importance of international cooperation and aligning national AI regulations with global frameworks to ensure consistent AI governance.

Similar Viewpoints

Both speakers emphasize the need for flexibility in AI governance approaches, allowing for adaptations as more countries join the treaty and as AI technology evolves.

AUDIENCE

NIGEL CASSIMIRE

Adopting a balanced and targeted approach to AI governance

Potential for future adjustments as new countries join

Unexpected Consensus

Importance of ongoing review and adaptation of the treaty

AUDIENCE

NIGEL CASSIMIRE

Ongoing review and updates through Conference of Parties

Potential for future adjustments as new countries join

Despite representing different perspectives (a government representative and an external observer), both speakers agree on the importance of allowing the treaty to evolve over time. This unexpected consensus highlights a shared recognition of the rapidly changing nature of AI technology and the need for governance frameworks to adapt accordingly.

Overall Assessment

Summary

The main areas of agreement include the importance of the AI treaty as a first legally binding international agreement, the need to balance innovation with safeguards, and the significance of international cooperation in AI governance. There is also consensus on the need for flexibility and ongoing adaptation of the treaty.

Consensus level

There appears to be a high level of consensus among the speakers on the fundamental principles and goals of the AI treaty. This strong agreement suggests a shared commitment to establishing effective global AI governance while respecting national differences. The consensus implies that the treaty has a strong foundation for implementation and future development, although specific details of national implementation may vary.

Different Viewpoints

Balance between innovation and regulation

Allison PETERS

AUDIENCE

Harmonizing various international approaches to AI

Adopting a balanced and targeted approach to AI governance

While both speakers emphasize the importance of balancing innovation and regulation, they differ in their approaches. Allison PETERS focuses on harmonizing international approaches, while the AUDIENCE speaker (representing the UK) advocates for a more targeted, sector-specific approach.

Unexpected Differences

Overall Assessment

summary

The main areas of disagreement revolve around the specific approaches to implementing the AI treaty and balancing innovation with regulation.

difference_level

The level of disagreement among the speakers is relatively low. Most speakers agree on the fundamental principles and goals of the AI treaty, with minor differences in implementation strategies. This suggests a generally unified approach to AI governance, which is likely to facilitate the treaty’s adoption and implementation across different countries.

Partial Agreements

Both speakers agree on the need to protect human rights while fostering AI innovation. However, they differ in their emphasis, with Allison PETERS focusing on establishing a shared baseline, while the AUDIENCE speaker (representing various countries) highlights the need for flexibility in national approaches.

Allison PETERS

AUDIENCE

Sets a shared baseline for rights-respecting AI development

Balances innovation and safeguards for human rights

Similar Viewpoints

Both speakers emphasize the need for flexibility in AI governance approaches, allowing for adaptations as more countries join the treaty and as AI technology evolves.

AUDIENCE

NIGEL CASSIMIRE

Adopting a balanced and targeted approach to AI governance

Potential for future adjustments as new countries join

Key Takeaways

The AI treaty is the first legally binding international agreement on AI, setting a shared baseline for rights-respecting AI development.

The treaty aims to balance innovation with safeguards for human rights, democracy, and rule of law.

It is open for participation from countries worldwide, not just Council of Europe members.

Countries are taking varied approaches to implementing the treaty and aligning it with national AI regulations.

International cooperation and multi-stakeholder input are crucial for effective AI governance.

The treaty is designed to be flexible and adaptable as AI technology evolves.

Resolutions and Action Items

Develop more detailed risk assessment frameworks for AI systems

Continue efforts to broaden participation in the treaty beyond current signatories

Work on aligning national AI regulations with the treaty’s principles

Establish a Conference of Parties for ongoing review and updates to the treaty

Unresolved Issues

Specific mechanisms for combating AI-driven disinformation while protecting free speech

Detailed implementation strategies for different national contexts

How to ensure the treaty remains relevant as AI technology rapidly advances

Balancing prescriptiveness of the treaty with flexibility for innovation

Suggested Compromises

Creating exemptions for research and development to allow innovation while maintaining safeguards

Adopting a general framework in the treaty while allowing for more detailed national regulations

Balancing clear legal obligations with flexibility for different regulatory approaches

Using a risk-based approach to determine the level of regulation needed for different AI applications

We all see the upsides of AI. We all see the benefits for development, for economic opportunity. But with every new phase of digital technology, we’ve seen human rights, the rights of women and girls, the rights of freedom of speech, democracy jeopardised.

speaker

Ambassador Dowling

reason

This comment succinctly captures the core tension at the heart of AI governance – balancing innovation and opportunity with protection of rights and democracy.

impact

It set the tone for much of the subsequent discussion, with many speakers referring back to this balance between innovation and safeguards.

We don’t want to do anything as it relates to regulation that cracks down on innovation, doesn’t allow that innovation to happen, and so certainly, with this convention, we see a convention that allows us to harmonize various different international approaches to AI, and one that still allows companies to innovate, to be creative, to create new AI systems that help really advance the opportunities as it relates to AI, but also help crack down on some of those risks that we’re talking about.

speaker

Allison Peters

reason

This comment articulates a key goal of the treaty – to create a harmonized international approach that protects rights while still enabling innovation.

impact

It shifted the discussion towards how the treaty aims to strike this balance in practice, leading to more detailed explanations of specific provisions.

It’s not the UN, but at the same time we felt it was extremely important to try to get as many member states from as many regions involved in the process. And so throughout the nearly two years of negotiation we saw a gradual increase not only in the number of participants but in fact signatories, hopefully at the end of the day.

speaker

Mr. Fairchild

reason

This highlights the importance of broad international participation, even for a treaty originating from a European body.

impact

It led to further discussion about how the treaty aims to be globally relevant and applicable beyond just Europe.

Turkey’s unique position, geopolitical and geostrategic position influences its approach to global AI governance. That’s why we have been actively monitoring and participating in international efforts.

speaker

Isil Denemec

reason

This comment introduces the perspective of how a country’s specific geopolitical context shapes its approach to AI governance.

impact

It broadened the discussion to consider how different national contexts might influence implementation of the treaty.

In the convention, there is a sort of, I think I touched upon it in my statement as well, but an exemption for research and development. So what that doesn’t mean is it just excludes all and every single form of research and development from the convention, but it creates a sort of a limitation and a safeguard.

speaker

Mr. Gibson

reason

This comment provides insight into a specific, important provision of the treaty that balances innovation with safeguards.

impact

It led to a more detailed discussion of how the treaty aims to practically balance innovation and protection in its specific provisions.

Overall Assessment

These key comments shaped the discussion by consistently returning to the central theme of balancing innovation with protection of rights and democracy. They highlighted the treaty’s aim to be globally relevant, adaptable to different national contexts, and practically implementable. The discussion evolved from broad principles to more specific provisions and implementation challenges, while maintaining a focus on international cooperation and harmonization of approaches.

How can the risk assessment framework for AI systems be developed and implemented across different countries?

speaker

Allison Peters

explanation

This is important for creating a shared baseline for assessing AI risks and implementing safeguards across different regulatory systems.

How can the convention contribute to fighting AI-driven disinformation and harmful content online?

speaker

Lina Viltrakiene

explanation

This is crucial given the alarming rise of disinformation that undermines human rights, particularly for vulnerable groups.

How can the convention be used to implement provisions from the Global Digital Compact (GDC)?

speaker

Online participant (unnamed)

explanation

Understanding this relationship is important for aligning international efforts on AI governance.

How can the convention be adapted or adjusted as additional countries join?

speaker

Nigel Cassimire

explanation

This is important for ensuring the convention remains relevant and inclusive as it expands beyond its initial signatories.

What impact will the treaty have on the daily lives of AI developers and coders?

speaker

Dr. Rafael de Lis Aguirre

explanation

Understanding the practical implications for the technical community is crucial for effective implementation of the treaty.

How can the convention balance addressing AI risks while safeguarding fundamental freedoms like free speech?

speaker

Ambassador Dowling

explanation

This balance is critical for ensuring the convention protects democracy without enabling censorship.