Science-Policy Interface Platform (UN MGCY) and YOUNGO (the official youth constituency to the UNFCCC)

1. Inclusive participation Participation of marginalised groups, including youth, is a prerequisite for the success of the Dialogue. Existing UN participation pathways ('UNPPs'), e.g. the UN MGCY in the case of a global youth demographic, should be used to nominate youth representatives to the Dialogue and to establish a pathway for youth to contribute to the work of the Independent International Scientific Panel on AI ('IISPAI'). For this to work, UNPPs must have their digital presence updated, with current representation and governance details publicly available across all UN member states. UNPPs should be directed to use national or regional legal frameworks of UN members to strengthen accountability mechanisms and thereby increase trust in UNPPs across regions and countries. The use of UNPPs mechanisms should be complemented by clear communications that inform individuals, and in particular marginalised groups, how they can use UNPPs to participate in the AI Governance Dialogue and contribute to the work of the IISPAI. These communications should be translated into all official UN languages and languages used by marginalised groups. Technical support and resources should be made accessible to marginalised groups to support their leadership in developing translation technologies. 2. Principles for AI governance identified The Dialogue should provide an opportunity for UN members to make representations and find consensus as to the ethical principles a global AI governance system should contain (the 'Overarching Principles'). The IISPAI should engage in the Dialogue to ensure its members understand the Overarching Principles for the purpose of devising Annual Reports that identify gaps in achieving the Overarching Principles and make recommendations for securing the practical implementation of the Overarching Principles. The Overarching Principles should be written in accessible language so that all people can understand them.

• Safe, secure and trustworthy AI
• AI capacity-building
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

An AI model that is able to find severe vulnerabilities in 'every major operating system and web browser' poses a threat and opportunity. It is claimed such a model exists. An opportunity is the utility such a model could provide in strengthening public infrastructure by means of enhanced vulnerability testing. A threat includes exposing public infrastructure to attack should similar models 'go rogue' or come into the possession of nefarious actors. Regardless of such a model's existence, belief that such a model exists sows fear and distrust amongst global society and thereby threatens global peace and security. The AI Dialogue should address potential safeguards against such threats. Without targeted capacity building, the advantages of AI technologies risk disproportionately accruing to those already in possession of advanced digital production technologies. This could exacerbate existing inequalities. Furthermore, indigenous communities remain largely excluded from AI benefits due to shortfalls in large language models. AI technologies bring enhanced capability for authoritarian practices and therefore threaten the spirit of the United Nations as espoused within its founding Charter. Articles 7, 12, 18, and 19 of the UDHR are under threat by AI surveillance techniques, internet scraping, and predictive policing. Such risks should be brought to the attention of a global public via the AI Dialogue as a pathway to democratic resilience and accountability in an age of AI. In harnessing AI technologies for medical purposes, a distinction between therapies and augmentation should be drawn since the realisation of the former may advance Article 25 of the UDHR, whilst the latter risks widening power imbalances as between human beings. Investment should be channelled towards medical AI therapies. Human oversight of AI technologies used in governance infrastructure is vital for honouring social contract theory. Mandating transparency on how AI is used by public bodies is therefore crucial.

Environmental concerns Rapid scaling of AI infrastructure is creating one of the fastest-growing sources of energy demand on the planet, with profound implications for national emissions trajectories and extractive practices that affect vulnerable nations most. An AI governance framework that does not consider environmental sustainability is incomplete. Peace and conflict The Dialogue presents an opportunity to moderate the AI arms race. States and private actors are increasingly deploying AI technologies in weapons systems, autonomous targeting, and battlefield decision-making with irreversible consequences that risk outpacing international humanitarian law. This issue cuts across safety, governance, and equity simultaneously, yet no single existing forum adequately bridges its technical, ethical, and geopolitical dimensions. At the same time, the AI arms race is diverting resources from humans to infrastructure. Whilst national authorities could devise regulations to moderate the amount of resources allocated to producing and powering AI technologies, the global AI arms race places pressure on governments to not fall behind or experience a fear of missing out. The AI Dialogue is a unique opportunity for competing countries to pause and slow the pace of reaching 'AI dominance'. Data protection Given the ostensible impossibility of deleting data used as AI training input, AI technologies that use personal data in machine learning processes pose a risk to data protection rights existing across many countries. Cutting edge technologies, including blockchain, which some research finds may facilitate the 'right to be forgotten' from AI models, should be explored, with expertise for developing such technologies identified and supported. An assessment of the right to erasure in the context of AI training should be carried out, with benefits (in particular with regard to medical AI technologies) and harms laid out to determine whether such right should cease to exist or become a minimum AI governance standard globally.

A lack of global AI governance impinges upon the principle of sovereign equality as between UN member states. Difficulties in developing AI technologies across Global South countries means a lack of empirical grounding needed to inform legislative drafting. This perpetuates a digital and regulatory gap, fomenting dependency on foreign-developed AI infrastructure and the importation of AI rules. Targeted support to Global South countries for developing ADP technologies is needed to reverse this dynamic. In the EU, a lack of AI governance legislation outside its jurisdiction is driving regulatory arbitrage. Instead of developing AI within the EU, entrepreneurs relocate to more permissive jurisdictions. EU governments must therefore depend upon AI technologies created by foreign technology firms that have scaled to the extent that they can absorb EU compliance costs. Establishing global minimum safety standards through international governance pathways could level the playing field and reduce talent flight without compromising AI safety protections. Morocco, the USA, and Brazil are experiencing conflicts in their management of crucial resources due to pressure to advance AI technologies. In Morocco, authorities have diverted water away from local populations to sustain data centres, exacerbating the existing water scarcity crisis. In Brazil, energy is documented to have been diverted towards powering data centres, thereby increasing the costs for ordinary citizens. In the USA water is being diverted to datacentre management during droughts. Weak enforcement of data protection laws combined with strong commercial incentives to harvest personal data for machine learning purposes has resulted in suspected unaccounted-for infringements of the right to privacy in Brazil. In India, a lack of algorithmic transparency is compromising accountability in AI deployment in hiring, finance, and public services. AI technologies demonstrate value in medical diagnostics, streamlining bureaucratic processes, and reducing waste.

Establishing a knowledge base Establishing inventories of reliable sources of data and knowledge can help advance implementation of the Overarching Principles (as defined in question 8). Such inventories can reduce the risk of satellite debates between UN members as to the existence of facts and/or the reliability of expert or academic opinions. We expect the International AI Safety Report to feature within this inventory. So as not to stifle its independence, we suggest the IISPAI (as defined in question 8 answer) be permitted to refer to any sources it deems fit in its Annual Reports, however, UN member states may use AI Governance Dialogues to express their dismissal of such sources as unreliable, and provide reasons why. Standardised terminology To clear up misconceptions as to what AI technologies are, a standardised dictionary of terms could be selected at the Dialogue. For example, given recent advancements in large language models, across some countries the term 'AI' is colloquially used to refer to LLMs specifically. A lack of appreciation of the greater breadth of AI technologies risks underestimating their risks and opportunities. Acknowledgement of contributions to the AI ecosystem Whilst dominant global AI companies originate from a few developed countries, the Dialogue presents an opportunity to acknowledge the crucial role that countries from which personal data and natural resources are extracted play in enabling global AI ecosystems. AI governance decision-making powers should be distributed between countries accordingly. Sharing of stakeholder perspectives By being given the opportunity to express themselves and their experiences, stakeholders can shed light on AI technologies to high-level delegations so that UN members can inform themselves as to appropriate next steps in terms of global AI governance. It is an opportunity for public-private cooperation across borders, suited to the porous nature of AI technologies.

International Organisation for Standardization (ISO) Despite geopolitical rivalry between AI leaders, patterns of competition and cooperation reveal that technical experts worldwide collaborate on the development of AI standards within ISO. As ISO standards are relatively uncontentious, in devising a technical AI governance framework, reference should be made to ISO standards. Technology Executive Committee (TEC) and the Climate Technology Centre Network (CTCN) Having accumulated research since 2016, the TEC and CTCN can help to bridge the cross-cutting issue of climate change and AI. The Dialogue should engage these bodies both as a mechanism for integrating climate considerations into AI governance in a technically grounded way and for addressing digital divides by drawing on technology needs assessments. Lex AI Bodies of soft law and voluntary norms on AI ('Lex AI') should be built upon. Lex AI can be discerned from, inter alia, the practices of AI startups and technology companies within themselves, as between themselves, and in their interactions with other stakeholders. The Asilomar AI Principles, signed by a broad coalition of academics, non-governmental organisations, founders, and international organisations, similarly reflect an emerging consensus of Lex AI. The AI Dialogue should engage with and build upon these frameworks, as they carry significant legitimacy and practical reach. Youth and Emerging Voices: ITU Young AI Leaders The ITU's Young AI Leaders programme offers a ready-made pipeline for next-generation AI expertise. The Dialogue should integrate this constituency to longstanding UN youth constituencies to allow for the sharing of information, expertise, capacity building, and to demarcate mandates. Grassroots organisations Civil society organisations focusing on digital rights, data worker representation, and community engagement should be invited to the AI Dialogue to share the challenges and solutions they have encountered in supporting AI adoption and deployment within their countries.

Establishment of working groups The IISPAI (as defined in question 8 answer) should identify the expertise needed to effect practical implementation of the Overarching Principles (as defined in question 8 answer) and establish working groups to deepen such expertise. Calls for experts to join the IISPAI working groups should be made by the Co-Chairs. The Secretariat with support of the UNPPs (as defined in question 8 answer) should shortlist working group applicants for selection by the IISPAI. To preserve the independence of the IISPAI, working group members should contribute in their individual capacity, with a mandatory declaration of disclosure of interests being a condition of participation. Side Events The Dialogue should establish a mechanism for side events through which updates from parallel governance processes and industry can feed into the Dialogue. Members of the Youth Science Policy Interface and relevant projects such as Project Glasswing should be invited to share details on relevant projects by way of Side Events. Information sharing and potential collaboration with the IISPAI should be considered, with potential for a member of the IISPAI to be included or have full visibility of Project Glasswing to report back on its progress at Dialogues. Shining a light on Project Glasswing and other similar projects is necessary for strengthening accountability of powerful AI actors. International Energy Agency The IEA could contribute knowledge and understanding on the constraints that shape how AI infrastructure can be built and deployed. Since energy infrastructure represents one of AI's most tangible leverage points, and the IEA has a unique dataset on compute concentration, the IEA's involvement could anchor equity debates with evidence and help UN members come to an AI governance consensus. It could further develop standardised energy efficiency benchmarks, giving the Dialogue concrete, technical metrics around which governments and industry can align.

Among voices underrepresented in global AI governance discussions are those of a global youth demographic who have little agency in AI governance but who disproportionately bear its consequences. Their interests should be actively taken account of by decision-makers at the AI Governance Dialogue. We ask that the Co-Chairs raise awareness of the UNPPs as pathways for youth and marginalised group participation in the Dialogue within the formal programme scheduled to be published this month (May 2026). Awareness and engagement campaigns offer a practical model for inclusion. The Swedish Embassy in Islamabad, together with the UN, demonstrated this by convening industry leaders and youth around sustainability themes, informing young people of ongoing initiatives and helping them engage with their national governments on policy matters. A comparable approach applied to AI governance could expand youth participation, particularly in regions where awareness of formal governance processes remains low. Civil society organisations have a critical role to play in amplifying youth voices and facilitating access to spaces that can otherwise be difficult to enter without institutional affiliation. Grassroots communities and people directly affected by AI systems, including gig workers, informal sector workers, and marginalised groups, are similarly excluded despite having the most immediate stake in governance outcomes. Non-technical professionals and local language users face comparable barriers. Organisers of the AI Dialogue can address this through proactive outreach. This could include targeted village-level consultations using local radio and community facilitators, materials translated into local languages including Luo, Swahili, and formats accessible to persons with disabilities such as Braille, and provision for virtual delegate quotas. Representatives of SMEs should be included to ensure the inclusion of their perspectives and to gauge how best to support them in adopting industrial norms, many of which have been set by powerful incumbent technology firms.

Fostering inclusive engagement in the AI Dialogue requires moving beyond traditional conference formats and embracing innovative, accessible, and participatory approaches. Transparency should be foundational, with regular information sharing through diverse media including blog writing, and publicly accessible summaries of progress on official UN websites. Live streaming sessions on platforms such as Tencent Video, VK Video, YouTube, could dramatically expand the reach of discussions to audiences who cannot attend in person. Encouraging the facilitation of convenings of Model United Nations adapted to the AI Governance Dialogue could secure youth engagement. By replicating the MUN structure around AI topics, young representatives from each country could simulate the mechanisms of the Dialogue, developing working knowledge of the process. Such an approach, however, should be cautiously adopted, given that MUNs are often costly to run and the cost is passed on to youth attendees, meaning inclusivity is impacted. Simulation exercises and gamified problem-solving (where participants role-play as delegations, companies, or affected communities navigating real governance dilemmas), hackathons can make complex regulatory questions tangible and engaging, particularly for non-technical audiences. Storytelling is another powerful tool. Community radio, local influencers, and village screenings could be engaged to translate abstract AI governance questions into everyday language, grounding global debates in the lived experiences of farmers, women, gig workers, and youth. Stories of how AI already affects people's lives make the stakes of governance decisions immediately legible. Offline submission portals and facilitator-led dialogues held in schools, markets, and faith spaces, to reach communities without reliable internet access, should be organised. All content, written, audio, and visual, should be made available in multiple languages, including indigenous languages, and communicated in accessible, non-technical terms to ensure no community is left out of the conversation.

The EU AI Act provides a regulatory model that is proportional in approach, requiring rigorous compliance for AI technologies that pose a risk to human rights. National AI strategies are emerging, including Zimbabwe's National AI Policy. Singapore's Model AI Governance Framework guides organisations towards safe and responsible AI deployment through practical implementation tools rather than prescriptive mandates. On intellectual property and human oversight, the Authors Guild's human-authored certification addresses concerns about AI-generated content displacing human creators, offering a market-based mechanism for protecting human authorship. This could be replicated across the arts and culture sectors. OpenAI's Industrial Policy for the Intelligence Age addresses (1) distributing AI-driven prosperity more evenly, (2) creating safeguards to reduce systemic risks, and (3) ensuring widespread access to AI so economic opportunity is not concentrated among a narrow technology class. It considers the fiscal implications of AI-driven economic transformation, including a transition away from income taxes towards capital gains and corporate taxation, a question of growing relevance for governments globally. The Youth Science-Policy Interface Platform is the recognised constituency for young scientists, engineers, practitioners, and students--from both the social and natural sciences--to contribute to the science-policy interface within the UN system. The SPI platform strengthens youth engagement in AI governance mechanisms and practice in sustainable development by equipping children and youth with tools to drive empirically-informed, context-specific, and fit-for-purpose change through science, technology, innovation, and data. The platform is managed by the UN Major Group for Children and Youth (UN MGCY). FunDoo, a UNICEF-partnered digital platform, demonstrates how inclusive governance can be built from the ground up. Through gamified peer learning, it teaches young people about AI literacy, ethics, and governance. UNESCO's Recommendation on the Ethics of Artificial Intelligence is a standard-setting instrument informing discussions on AI governance and implementation of the Global Digital Compact.