Cooperative AI Foundation

The first Global Dialogue will have succeeded if it translates the UN's convening legitimacy, in which all Member States participate equally, into visible, inclusive progress on AI governance that no unilateral or regional process can replicate. AI development today is shaped by powerful forces: optimization for user engagement, competition to capture value from automation, and divergent national and commercial approaches. A successful Dialogue would demonstrate that Member States, acting together, can direct this momentum toward human well-being, equitable benefit-sharing, and the protection of human rights and fundamental freedoms, ensuring that the benefits of AI are broadly shared rather than concentrated among a few. Several indicators would signal success: One would be a shared articulation of the values underpinning AI governance, consistent with UN instruments: human agency; shared prosperity and equitable access; human dignity; human rights and self-determination; and responsibility and accountability. Safe, secure and trustworthy AI systems make these values achievable, supported by a shared understanding of the risks of common concern and of foundational safety principles that can guide AI development globally. Another would be convergence on governance itself, the Dialogue's mandate. This would look like Member States exchanging frameworks and regulatory experience across regions, surfacing common principles, and identifying practical avenues for cooperation that build on existing commitments. Useful signals would include convergence around governance approaches, shared risk taxonomies, regulatory cooperation, capacity-building, access to AI's benefits, bridging digital divides, and safeguards for human rights and fundamental freedoms. A further indicator would be an accountable path forward which could include a manageable set of follow-up workstreams, regionally balanced co-leadership, a clear reporting cycle, and a defined route for stakeholders to inform future deliberations.

• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Safe, secure and trustworthy AI

While all seven of the identified thematic areas warrant urgent attention, we're best positioned to contribute meaningfully to these four areas. Safe, secure and trustworthy AI: Human well-being is enabled through AI systems that we can understand, control and that enhance rather than harm. Safety and security are the prerequisite. When States establish a shared understanding of unacceptable risks, safety becomes a competitive advantage, preventing fragmentation and enabling trustworthy development across all countries. The interoperability and compatibility of artificial intelligence governance approaches: As AI governance efforts multiply, fragmentation increases. National, regional, and international initiatives are emerging rapidly, often without clear paths to coordination. With effective coordination, differing approaches can be made complementary, providing coherence for governments, companies, regulators, civil society and the public. The Dialogue can serve as a space to share experiences and develop governance approaches that support coordination across jurisdictions and help ensure meaningful protections for people, while respecting the diversity of national and regional frameworks. Additionally, the Dialogue should underscore that effective assurance and verification infrastructure is needed to give AI governance approaches practical effect. Protection and promotion of human rights: Safe, secure and trustworthy AI cannot be achieved through technical measures alone. Human rights protections are integral to it. Building on longstanding international commitments such as the Universal Declaration of Human Rights, AI development must respect, protect and fulfill civil and political, as well as economic, social and cultural rights, advancing human agency, dignity and self-determination. These rights are foundational to AI governance, grounding its development. Transparency, accountability and human oversight: Credible governance empowers all States to understand AI systems and decisions, hold actors accountable for harms and ensure that humans retain meaningful control and oversight over AI models. Through transparency and peer learning, safety standards can be verified, capacity- building outcomes can be audited and human rights can be protected globally.

While the listed themes provide a comprehensive framework, we draw attention to the importance of addressing risks arising from advanced AI systems operating in the real world, which can be understood as multi-agent environments. Addressing emerging multi-agent AI risks: As AI systems become increasingly autonomous and interconnected, ensuring their safe deployment requires moving beyond evaluations conducted purely in isolation. Research by the Cooperative AI Foundation on multi-agent risks highlights the urgent need for robust and scalable evaluation methods that can assess advanced AI agents in realistic multi-agent settings, where coordination failures, strategic misalignment, collusion, and other emergent behaviours may create novel safety and security challenges. Current evaluation approaches remain limited, often focusing on single-agent performance or narrow multi-agent settings that do not adequately reflect real-world deployment conditions. The International AI Safety Report 2026 recognises multi-agent risks as an emerging area of concern, reflecting growing consensus that these dynamics warrant dedicated international attention. We therefore hope that the Independent International Scientific Panel on AI will pay particular attention to multi-agent risks in its future work, including identifying research gaps, evaluation standards, and governance challenges associated with increasingly networked and autonomous AI systems. The UN is uniquely positioned to facilitate international dialogue on these issues, particularly where risks are cross-border in nature and may affect critical infrastructure, economic stability, and information ecosystems.

The UN Global Dialogue can advance international cooperation on AI governance in a way that complements other international efforts, drawing on the convening power of a universal forum. It reinforces shared values that steer AI development to serve humanity and provides a reference point for governance efforts at national, regional and international levels. It promotes human rights-grounded governance that can drive transparency, accountability, and meaningful human oversight in AI systems. Member States can respond coherently rather than in isolation, reducing risk of fragmentation when serious risks emerge. The Dialogue can also enable Member States to learn from one another's experience: which governance approaches are working, where implementation challenges arise, where divergences are emerging. This transparency supports voluntary convergence and identifies gaps requiring collective attention. One such gap is the absence of a shared approach to prevention. From disaster risk reduction to public health, international experience shows that early action is more effective and less costly than response after harms have materialized. This is especially salient for AI, where certain risks, whether to essential services, public health systems, or information integrity, may be irreversible and cross borders rapidly. The Dialogue can create space for exchanging on preventive approaches that remain attentive to the capacity constraints and development priorities of all Member States. Moving from shared frameworks to meaningful implementation requires capacity. Without sustained support for AI governance capacity building, particularly for States that lack AI Safety Institutes or comparable national bodies, shared frameworks risk uneven implementation with the burden of harm falling on States least equipped to detect and respond.

International governance frameworks: UN Global Digital Compact; UNESCO Recommendation on the Ethics of AI; UN Global Principles for Information Integrity, and the UN Guiding Principles on Business and Human Rights; ASEAN Guide on AI Governance and Ethics; Hiroshima AI Process; Bletchley, Seoul, Paris, and Delhi AI Summit outcomes; AU Continental AI Strategy; EU AI Act and other pieces of national legislation or strategies; Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law; OECD AI Principles and OECD-integrated Global Partnership on AI (GPAI); Santiago/Montevideo Declarations on the Ethics of AI in Latin America and the Caribbean. (See OECD.AI Policy Navigator and the Center for AI and Digital Policy's "AI Policy Sourcebook" for overviews of AI policies and practices worldwide.) Technical standards: ITU's AI technical recommendations; CEN CENELEC and ISO/IEC standards; relevant IEEE frameworks. Implementation and capacity building networks: Regional development bodies and capacity building initiatives; emerging national AI authorities, including AI safety institutes. Research and expert networks: Independent International Scientific Panel on AI, the International AI Safety Report, the Singapore Consensus on AI Safety Research Priorities, the International Association for Safe and Ethical AI (IASEAI), Partnership on AI (PAI), the Global South Network for Trustworthy AI. The UN Global Dialogue can offer a space in which to coordinate fragmented technical and regional initiatives, create transparency about alignment and gaps, enable political backing for implementation, and foster a forum for peer learning.

Brazil: rights-based approach through PL 2338/2023 (pending enactment); prohibits excessive- risk AI; establishes rights to explainability and human review. China: comprehensive regulatory approach through PIPL (2021), Algorithm Recommendation Provisions (2022), and Interim Measures on Generative AI (2023); mandates algorithmic transparency, content moderation and security assessment; actively advancing international governance through Global AI Governance Initiative. EU: legally binding risk-based governance through EU AI Act; prohibits "unacceptable risk" practices including manipulation causing harm, public-sector social scoring and certain biometric identification; mandates conformity assessments, post-market monitoring and human oversight. India: "techno-legal" approach combining baseline data protection safeguards, sectoral regulation and technical controls; emphasizes inclusion and digital public infrastructure through IndiaAI Mission rather than single overarching AI law. Singapore: innovation-enabling approach through Model AI Governance Framework (including 2024 Generative AI guidance) and open-source AI Verify testing framework; emphasizes voluntary adoption and interoperability. South Africa: human-centric, ethics-first approach through National AI Policy Framework (2024), emphasizing Ubuntu principles, inclusion and alignment with AU Continental AI Strategy. South Korea: innovation-led approach through AI Basic Act (2026) with risk-based obligations for "high-impact AI", active international coordination through AI Safety Institute and Seoul Summit. UAE: innovation-first positioning through the National AI Strategy 2031 and dedicated AI Minister (since 2017); substantial investment in domestic capacity and active international AI engagement. USA: innovation-first approach without federal omnibus AI legislation; combines voluntary NIST AI Risk Management Framework, targeted federal statutes (TAKE IT DOWN Act) and active state legislation (California SB 53, Texas Responsible AI Governance Act, New York RAISE Act, Virginia SB 384 / HB 797). Using key building blocks of AI governance as a comparative framework, the Dialogue can identify which are working in practice, where capacity gaps persist and which safeguards should become interoperable across jurisdictions: https://global-governance.ai/.