IGF Daily Summary
for Thursday, 26 June 2025
Dear readers,
Welcome to our daily report from Day 3 at IGF 2025.
On the third day, the IGF discussed its future ahead of WSIS+20 negotiations. It is happening in the broader context of an unprecedented crisis at the UN and UN80 debates about the future of organisations. There is no enthusiasm for new bodies and forums in such a context, making the IGF’s future role even more critical.
While there is a shared opinion that the IGF should continue to evolve, discussants expressed a wide range of views about concrete steps and reforms. The Leadership Panel argued for stronger institutionalisation of the IGF, while some discussants warned that such a move could reduce the Forum’s flexibility and agility.
Another issue that stood out for us in yesterdays’ discussions was that digital sovereignty is becoming a priority for developing countries instead of the traditional focus on connectivity and infrastructure. For example, they aim to become content generators instead of just consumers of content provided somewhere else.
The third day debates also featured emerging technologies and their governance challenges, from open-source AI applications to autonomous weapons systems. Participants discussed how emerging technologies are reshaping power dynamics between states, private entities, and users, while exposing vulnerabilities in critical infrastructure.
What stood out to you in yesterday’s discussions?
Diplo reporting team
Key questions from Day 3
1. How can child safety be ensured in algorithmic environments?
Participants emphasised that self-regulation has not worked. Instead, regulation is needed, such as the Digital Services Act. Key recommendations included privacy and safety by design and default, turning things off to ensure children’s experience stays private with real agency and choice, and implementing age-appropriate design principles. Participants noted that platforms need to take more responsibility for taking down content that is damaging and prohibited, and secure appropriate age verification.
2. What should be the IGF’s role in digital governance?
Participants strongly supported the renewal of the IGF with adequate resourcing, describing it as the ‘only place globally where stakeholders can come together as peers’. The IGF was described as a successful prototype for implementing the multistakeholder approach in the UN system, which could be built upon to strengthen multistakeholderism in other parts of the UN system. Some concerns were expressed ‘about the IGF trying to juggle everything and please everyone,’ which was described as a disservice because it makes it impossible to dive deeply into particular topics.
3. How can AI divides be bridged sustainably?
Participants noted that ‘between 2022 and 2025, AI-related investment doubled from $100 to $200 billion’, which is about three times the global spending on climate change adaptation. Participants suggested transitioning from brute force models that are large and energy-consuming to vertical and agile foundation models with specific purposes. TinyML was presented as involving running machine learning models on extremely small devices, with the advantages of being low power and low cost.
4. How can AI development serve global needs equitably?
Participants noted that current AI deployment tends to reinforce existing inequalities while marginalising non-Western worldviews and indigenous knowledge. Participants suggested creating a CERN-like model for AI that could help pool resources to provide shared infrastructure for every stakeholder. The importance of working with community-led data and indigenous knowledge to focus on specific problems in local contexts was emphasised.
5. How can critical internet infrastructure be protected during conflicts?
Participants suggested creating multistakeholder mechanisms involving states, the technical community, private telecommunications companies, and civil society to respond to crises, providing funding and political leverage to allow repair equipment into affected areas. Technical solutions such as mesh technologies should be enforced into phones, allowing devices to communicate with each other for emergencies, particularly for low-bandwidth text communications
Summary of discussions
AI governance and development
Participants noted that policy responses to AI are very often fragmented, reactive and dominated by short-term interests, with no continuity or globality in addressing policy responses to AI’s impact. They also highlighted that current AI deployment tends to reinforce existing inequalities while marginalising non-Western worldviews and indigenous knowledge.
Discussions also revolved around the concentration of AI power, with participants noting that the real power of AI is concentrated in a few companies in a few countries. ‘Balancing the speed of how this technology is evolving (…) with the depth of the safeguards that we have to provide’ was identified as a primary challenge.
Environmental sustainability needs to be more prominent in AI governance debates. Sustainable AI will not emerge by default and needs to be actively supported and incentivised. TinyML was presented as an approach involving running machine learning models on extremely small devices with few kilobytes of memory and slow processors, but with the advantages of being extremely low power and low cost, with devices costing less than a dollar for chips and about $10 for full devices.
In the African context, open-source AI offers entrepreneurs, NGOs, governments, researchers, and academics affordable access to advanced tools that would otherwise require costly proprietary licenses. Yet over half the population of the continent are not connected to the internet, and there is a dearth of access to data and computing power for data scientists.
Child safety and digital platforms
The high-level session on child safety provided concerning statistics about children’s digital experiences. Consistently, around half of the children surveyed say they feel addicted to the internet, with rates of ADHD, depression, eating disorders, child sexual abuse, and suicide going through the roof. Nearly two-thirds say they often or sometimes feel unsafe online, with more than three-quarters encountering content they find disturbing, including sexual content, violence, and hate.
Participants noted that most services where children spend time are designed with three primary purposes geared towards revenue generation: maximise time spent, maximise reach, and maximise activity. This creates environments ‘where children can go from a simple search for slime to porn in just a single click, or from trampolining to pro-anorexia in just three clicks, and nudge to self-harm in 15 clicks’.
Research on children’s perspectives provided insights about AI development. When children learned about environmental impacts, particularly water consumption and the carbon footprint of generative AI models, they often chose not to use those models in the future. Children also identified that AI models consistently produced images of people that were white and predominantly male by default, causing distress, particularly for children of colour who felt unrepresented.
Effective regulation – illustrated by the Digital Services Act – often proves more successful than self-regulation, it was said. Recommended actions include building in privacy and safety by design and by default so children’s experiences remain private while preserving real agency and choice, and adopting age-appropriate design principles.
Digital infrastructure and connectivity
Digital exclusion translates into a ‘lack of access to essential services’, limiting opportunities and deepening inequalities. The problem is now less and less a coverage gap and more and more a usage gap requiring different interventions to close the gap, related to digital literacy, technology affordability, online safety, and ensuring that the online environment is compatible with linguistic and cultural diversity.
Community networks were discussed as representing a form of digital sovereignty, allowing communities to control and own their digital infrastructure while deciding how it is built, maintained and who benefits. However, one study showed that only a minority of surveyed initiatives cover all costs with autonomous revenues, with most requiring blended financing approaches combining private fees, community financing, public grants, and local government support.
The Arctic region presents unique connectivity challenges, with participants noting that ‘there are no roads, and there are no railways which would lead to these regions and people’. Solutions included creating IT camps where people, especially indigenous communities with nomadic lifestyles, can access free internet and digital resources upon arrival.
Cybersecurity and critical infrastructure
Cybersecurity sessions addressed evolving threats and governance challenges. Participants noted that ‘cybercriminals operate at the speed of light, while law enforcement or The Good Guys operate at the speed of the law, implying that it will always be a game of catch-up.’
Critical infrastructure protection was addressed through the lens of conflict situations. The session on securing internet access during crises noted that ‘armed conflict has become the leading trigger of internet shutdowns worldwide. What we see is that in times of conflict, warring parties see civilian internet infrastructure as a military target.[…] the warring parties weaponise access, such as banning or restricting access to repair parts, spare parts or critical components of the infrastructure that is needed to repair it.’
While the physical internet infrastructure falls within national borders, its cross-border nature raises questions about due diligence responsibilities and the prevention of transboundary harm.
Participants suggested creating a multistakeholder mechanism involving states, the technical community, private telecommunications companies, and civil society to respond to crises, providing funding and political leverage to allow repair equipment into affected areas.
Multilingualism and cultural diversity
Language barriers are obstacles to digital inclusion. With over 7,000 languages spoken worldwide, the dominance of a few major languages online prevents billions of users from fully participating in digital spaces. Almost 50% of online content is still in English, despite English not being the native, second, or third language for many internet users and those yet to come online.
The IDN World Report 2025 provided statistics showing roughly 70% of the estimated 4.4 million IDNs worldwide are under country code TLDs, but with almost minus 1% yearly growth for ccTLDs and even 5.5% growth for gTLDs. Technical implementation gaps were noted, as over half of registries do not support Unicode addresses in email servers at all, three-fourths do not permit Unicode symbols as contact emails in their registry database, and none of the ccTLDs stated they would offer support to internationalised email addresses.
A paradigm shift to multilingual first as opposed to English first was recommended, implementing multilingual by design and universal acceptance by design.
Environmental sustainability and e-waste
According to the Global E-waste Monitor, over 62 million tons of e-waste are generated annually globally, with only 32% formally collected and recycled, meaning 78% remains unmanaged and results in polluting land, air, and water while causing severe health risks. Data-driven technologies (AI, IoT, digital product passports) are crucial for making e-waste traceable and manageable – ‘we can’t manage what we can’t measure’.
Participants recommended implementing comprehensive regulations and policies focusing on strengthening and enforcing extended producer responsibility (EPR) with clear mandates that legally require manufacturers, importers and retailers to take responsibility. Moreover, environmental sustainability must be embedded in all business operations, with every digital activity having a measurable carbon footprint.
SDGs in focus
Several sessions made explicit connections between digital governance and the SDGs.
SDG 15 (Life on land) was referenced through Microsoft’s Project Sparrow, an AI-enabled device for environmental monitoring that supports SDG 15 by tracking biodiversity and habitat health in remote areas like the Amazon rainforest.
SDG 11 (Sustainable cities and communities) featured in the discussion on local digital governance. Participants noted that the Local Online Service Index (LOSI) network plays a critical role in supporting the achievement of SDG 11, making cities inclusive, safe, resilient and sustainable. Over 65% of SDG targets fall under the jurisdiction of local authorities, making local digital governance essential rather than just relevant.
SDG 5 (Gender equality) was referenced in connectivity discussions. While most African countries have adopted gender-equality frameworks, the missing piece is rigorous implementation: building gender-intentional digital infrastructure, systematically measuring and tracking its impact, and feeding those insights back into policy so gaps close sustainably and no one is left behind.
Participants emphasised that achieving the SDGs can only be done by unlocking opportunities through digital transformation. TinyML applications were noted as having an impact on SDGs, with diverse applications including disease detection in livestock, bee counting, anaemia detection, and wildlife behaviour monitoring.
Thought-provoking ideas and questions
‘Why won’t adults stand up for children? You watch everything we do online, you nag us to get off our devices, even though you stay firmly glued to yours, and now you just want to outright ban us. When are you going to stop making out that we are the problem instead of the system? Why don’t you stand up for us?’
‘AI doesn’t happen to us. The current narrative is often that AI is something like the weather. We have no idea how to control it. […] But AI is not weather. AI is developed by us, is developed by organisations, by people.’
‘Now the debate is not humans versus machines. Now the debate is about who understands and uses managed AI versus who doesn’t.’
‘Nobody’s coming to save us. We need to start thinking of ways where we can invest, locally invest in natural language processing, so that we can then call the shots.’
WSIS+20 review process and the IGF
The importance of inclusive and accountable multistakeholder participation was underscored throughout the Forum. Many participants pointed to the São Paulo principles as a valuable reference for fostering transparent and equitable engagement across WSIS, IGF, and GDC discussions. They stressed the need for clear accountability mechanisms to ensure stakeholder inputs are meaningfully considered and followed up on.
Looking ahead to the WSIS+20 review, it was noted that processes must be transparent, comprehensible, actionable, and accessible to diverse stakeholders. National and Regional IGFs (NRIs) should be considered special assets within discussions on the WSIS+20 review and IGF mandate renewal. NRIs are uniquely positioned to engage with local governments and broader communities for implementing GDC and WSIS outcomes. There was strong support for positioning the IGF as a central space for tracking the implementation of outcomes from both the GDC and WSIS.
Concerns were raised about overlapping UN processes, with calls for more streamlined and simplified approaches. Participants observed the potential value of aligning GDC implementation with the WSIS architecture to foster coherence and reduce duplication in digital cooperation efforts. It was suggested that UNGIS could consider integrating GDC priorities into existing WSIS action lines and explore joint implementation mapping. accountable’.
The IGF we want
Permanent mandate and enhanced institutional resourcing: Multiple participants emphasised the need for a permanent mandate for the IGF, along with support for national and regional IGFs to create a holistic and effective ecosystem. There were also recommendations for securing a long-term and stable financial foundation for the IGF to ensure the full implementation of its mandate and to strengthen the IGF Secretariat.
Cross-sectoral collaboration expansion: Participants suggested that the IGF should expand by extending invitations to educational ministries, finance ministries, and experts from other areas less represented at the Forum, as digital has become an enabler across all sectors, and decisions cannot be made in isolation.
Streamlined focus and priority setting: There were calls for greater prioritisation and streamlining IGF processes and intersessional work, to give the Forum more focus.
Enhanced accessibility and language support: A suggestion was made that there should be more possibilities to exchange in different languages at the IGF.
Diplo/GIP at IGF2025
Diplo is partnering with the IGF Secretariat and the Government of Norway (as host country) to deliver AI-enabled, just-in-time reporting from the IGF 2025 meeting. Building on a decade of just-in-time IGF reporting, we will continue to provide timely and comprehensive coverage from the forum. Our reporting initiative will include session reports, an ‘Ask IGF 2025’ AI assistant, daily highlights, and more, available on our dedicated IGF 2025 web page on the Digital Watch Observatory.
Yesterday, 26 June, the Diplo-led CADE consortium held an in-person lightning session exploring multilingual challenges in global processes. A ‘Tower of Babel’ simulation invited participants to speak their native languages, followed by a discussion on practical solutions for language barriers.
Still in Lillestrøm for IGF? We’re here too! Come visit the Diplo and GIP booth (#45) and the CADE booth (#57) — and don’t forget, both are also live in the virtual village.
Do you like what you’re reading? Bookmark us at https://dig.watch/event/internet-governance-forum-2025 and tweet us @DigWatchWorld
Have you heard something new during the discussions, but we’ve missed it? Send us your suggestions at digitalwatch@diplomacy.edu.