UNU-CRIS

A successful first Global Dialogue on AI Governance would deliver three concrete outcomes: shared direction, practical cooperation, and sustained inclusion. It should produce a common baseline of principles that are not merely aspirational, but operational. This includes convergence on risk management, transparency, human oversight, and accountability, with enough flexibility to reflect diverse legal systems while ensuring interoperability. Early alignment, particularly between regions such as the Indo-Pacific and the European Union, would reduce fragmentation and help prevent regulatory arbitrage. Success would mean launching actionable cooperation mechanisms. This could take the form of joint pilot projects on AI safety, shared protocols for addressing misinformation and disinformation (especially during elections and crises), and coordinated capacity-building initiatives. Given the uneven global distribution of AI capabilities, tangible commitments to support developing countries—through access to compute, data, and skills—would be critical to bridging AI divides. The Dialogue must demonstrate credible multi-stakeholder engagement. Governments alone cannot govern AI effectively. Meaningful participation from academia, the private sector, and civil society should be embedded in both agenda-setting and follow-up processes. In particular, integrating insights from those working on democratic resilience, media literacy, and collective decision-making can strengthen the societal foundations of AI governance. Finally, success would be measured by continuity. Establishing a roadmap with clear milestones, review mechanisms, and feedback loops would ensure that the Dialogue evolves alongside the technology it seeks to govern. In sum, the Dialogue should move beyond discussion toward coordination—grounded in shared principles, enabled by practical cooperation, and sustained through inclusive governance.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches

Advancing safe, secure and trustworthy AI is foundational. As AI systems are increasingly embedded in critical infrastructures and information ecosystems, the risks associated with misuse, systemic bias, and large-scale misinformation demand coordinated mitigation strategies. Priority should be given to developing practical risk management frameworks, shared safety standards, and mechanisms for rapid information-sharing during crises, including elections and natural disasters. Second, AI capacity-building is essential to ensure that no country is left behind. Current disparities in access to compute, data, and technical expertise risk widening global inequalities and limiting meaningful participation in governance discussions. Efforts should focus on strengthening institutional capabilities, supporting education and media literacy in the age of generative AI, and enabling developing countries to both adopt and shape AI technologies in line with their societal needs. In addition, promoting interoperability of governance approaches is critical to avoid fragmentation. Divergent regulatory frameworks across regions can create uncertainty, reduce accountability, and hinder effective cross-border responses to AI-related risks. Greater alignment, while respecting different legal and cultural contexts, can facilitate cooperation, enhance compliance, and support the development of globally consistent safeguards. These priorities are mutually reinforcing: trustworthy AI cannot be achieved without broad-based capacity, and interoperability depends on a shared foundation of capabilities and principles. Focusing on these areas will help translate high-level commitments into coordinated, practical outcomes.

Yes. While the proposed themes are comprehensive, several cross-cutting and emerging issues merit more explicit attention. Information integrity in the age of generative AI cuts across safety, human rights, and societal impacts. The rapid evolution of synthetic media is reshaping how information is produced, distributed, and trusted-particularly during elections, crises, and conflicts. This calls for coordinated approaches that combine technical safeguards, platform accountability, and updated media and information literacy frameworks. Collective decision-making under AI-augmented environments is an emerging governance challenge. AI systems increasingly influence not only individual choices but also group dynamics, public opinion formation, and policy processes. Addressing imperfections in democratic decision-making, such as information asymmetries and coordination failures, requires integrating insights from behavioral science, digital governance, and institutional design. Evaluation and measurement gaps remain underdeveloped. There is a need for shared methodologies to assess AI risks, societal impacts, and the effectiveness of governance interventions. Without comparable metrics, it is difficult to ensure accountability or learn across jurisdictions. Crisis governance and resilience should be highlighted. AI can both exacerbate and help manage crises, including natural disasters and security incidents. Mechanisms for rapid coordination, trusted information-sharing, and pre-agreed response protocols are essential. Finally, policy integration across domains is critical. AI governance cannot be siloed; it intersects with security, economic policy, education, and foreign policy. Strengthening coherence across these areas will be key to effective and adaptive governance. Addressing these cross-cutting issues would enhance the Dialogue's ability to respond to real-world complexities and evolving risks.

Governance gaps in "safe, secure and trustworthy AI", "AI capacity-building", and "interoperability of governance approaches" are already having tangible effects across Japan, the Indo-Pacific, and cooperation with European partners. One of the most significant challenges lies in **information integrity and crisis response**. The rapid diffusion of generative AI tools has amplified risks of misinformation and disinformation during elections, natural disasters, and security incidents. In the absence of harmonized standards for platform responsibility, verification, and cross-border information-sharing, responses remain fragmented. This undermines public trust and complicates timely, coordinated crisis management. A second challenge concerns capacity asymmetries. While advanced economies are deploying AI across public and private sectors, gaps in technical expertise, computational infrastructure, and institutional readiness persist both within and across countries. These disparities limit effective participation in governance processes and hinder the adoption of safeguards, particularly in parts of the Indo-Pacific. Without targeted capacity-building, there is a risk of widening both economic and governance divides. Third, regulatory fragmentation across jurisdictions, particularly between major frameworks in Europe and the Indo-Pacific, creates uncertainty for policymakers and industry alike. Divergent approaches to risk classification, data governance, and accountability can impede collaboration and reduce the effectiveness of cross-border risk mitigation efforts. At the same time, these gaps present important opportunities. There is growing momentum for inter-regional cooperation, particularly between Japan and European partners, to align governance principles and develop interoperable frameworks. Advances in AI also enable improved "early warning systems, multilingual communication, and decision-support tools" for crisis management. Furthermore, renewed emphasis on "media and information literacy" offers a pathway to strengthen democratic resilience against AI-enabled manipulation. Addressing these challenges through coordinated governance, capacity-building, and shared standards can turn current vulnerabilities into foundations for more resilient and inclusive AI ecosystems.

The AI Dialogue can play a pivotal role as a convening, aligning, and operational platform for international cooperation on AI governance. It can serve as a "neutral forum for convergence". By bringing together all Member States and stakeholders, the Dialogue can help identify common ground on core principles, such as safety, accountability, and human oversight, while respecting diverse legal and cultural contexts. This is particularly important for fostering interoperability between regional frameworks, including those emerging in the Indo-Pacific and Europe, and reducing regulatory fragmentation. Second, the Dialogue can move beyond principles to "enable practical cooperation". It can incubate joint initiatives such as shared risk assessment methodologies, coordinated responses to AI-enabled misinformation and disinformation, and collaborative research on AI safety. Establishing channels for real-time information-sharing during crises, such as elections or natural disasters, would be especially valuable. It can act as a hub for capacity-building and inclusion. By leveraging existing UN mechanisms and partnerships, the Dialogue can help mobilize resources, expertise, and infrastructure to support developing countries. Ensuring broad participation is essential not only for equity, but also for the legitimacy and effectiveness of global AI governance. The Dialogue can strengthen multi-stakeholder engagement. Structured participation from academia, the private sector, and civil society can enrich policy discussions and support implementation. In particular, incorporating perspectives on democratic resilience, media literacy, and collective decision-making can help address the societal dimensions of AI governance. The Dialogue can provide continuity and accountability by establishing follow-up mechanisms, benchmarks, and review processes. In a rapidly evolving technological landscape, sustained coordination is critical. In sum, the AI Dialogue can transform fragmented efforts into coherent global action, bridging regions, sectors, and capacities to advance inclusive and effective AI governance.

The AI Dialogue should build on and connect with a range of existing international, regional, and multi-stakeholder initiatives to avoid duplication and accelerate progress. At the global level, frameworks such as the United Nations system initiatives, including the Global Digital Compact, as well as the UNESCO Recommendation on the Ethics of AI, provide normative foundations. The OECD AI Principles and the work of the G7, including the Hiroshima AI Process, offer policy alignment among advanced economies. In addition, the Global Partnership on AI has developed valuable multi-stakeholder research and policy guidance, while the International Telecommunication Union contributes technical standards and capacity-building efforts. Regionally, the European Union's regulatory framework and Indo-Pacific cooperation initiatives provide important testbeds for governance approaches. Cross-regional dialogues and bilateral partnerships have also advanced collaboration on AI safety, data governance, and digital resilience. The added value of the AI Dialogue lies in its universality and integrative function. Unlike many existing initiatives with limited membership, it provides a platform where all countries, particularly developing economies, can participate on equal footing. It can act as a bridge across fragmented efforts, promoting interoperability between frameworks and facilitating mutual learning across regions. Moreover, the Dialogue can focus on operational coordination: aligning standards, supporting joint pilot projects, and enabling real-time information-sharing in areas such as AI risk management and responses to misinformation. It can also strengthen capacity-building synergies by mapping existing initiatives and directing resources where gaps are greatest. In essence, the AI Dialogue's value is not to replace existing mechanisms, but to connect, amplify, and operationalize them into a more coherent and inclusive global governance ecosystem.

Inclusive participation will be essential for the AI Dialogue to produce legitimate and actionable outcomes. Different stakeholders can contribute in complementary ways. Governments should provide policy direction, share regulatory experiences, and commit to aligning national frameworks where feasible. Private sector actors can contribute technical expertise, risk assessments, and insights on implementation challenges, particularly in deploying safe and trustworthy AI systems. Academia and the technical community can offer independent research, evaluation methodologies, and foresight on emerging risks. Civil society organizations play a critical role in representing societal interests, including human rights, democratic resilience, and media and information literacy. International and regional organizations can help coordinate efforts, scale capacity-building, and connect existing initiatives. To make these contributions effective, the Dialogue should adopt a structured yet flexible format. It should be organized around thematic working groups aligned with priority areas such as safety, capacity-building, and interoperability. These groups should be multi-stakeholder by design and tasked with producing concrete outputs, such as guidelines, toolkits, or pilot initiatives. Second, the Dialogue should include regular plenary sessions to review progress, ensure political buy-in, and maintain strategic coherence. Between plenaries, iterative consultation processes, including public calls for input and regional workshops, can broaden participation, particularly from underrepresented regions. Establishing practical cooperation tracks would enhance impact. These could include joint simulations (e.g., crisis response to AI-enabled misinformation), shared repositories of best practices, and collaborative capacity-building programs. The Dialogue should ensure continuity and accountability through clear timelines, measurable outputs, and transparent reporting mechanisms. A well-designed structure, combining inclusiveness with operational focus, will allow diverse stakeholders not only to be heard, but to actively shape and implement global AI governance.

Several voices remain underrepresented in global AI governance discussions, limiting both legitimacy and effectiveness. Developing and emerging economies, particularly from the Global South, are often insufficiently represented. This is not only a question of access, but of capacity—many countries face constraints in technical expertise, data infrastructure, and policy resources. As a result, their perspectives on development priorities, local risks, and cultural contexts are not adequately reflected. In addition, frontline practitioners in democratic processes and crisis response, such as election administrators, local governments, and emergency management officials are rarely included, despite being directly affected by AI-enabled misinformation and operational challenges. Their practical experience is essential for designing workable governance mechanisms. Educators and media literacy experts remain underrepresented, even though long-term societal resilience increasingly depends on how individuals understand and engage with AI-generated information. Similarly, civil society actors from non-Western contexts often lack meaningful opportunities to shape global norms. Small and medium-sized enterprises (SMEs) and open-source communities are frequently overlooked, despite their growing role in AI development and deployment. To address these gaps, the AI Dialogue should prioritize capacity-supported participation, including funding mechanisms, technical assistance, and preparatory briefings to enable informed engagement. Regional consultation platforms can help surface context-specific perspectives and feed them into global discussions. In addition, adopting hybrid and multilingual formats would lower barriers to participation. Importantly, inclusion should go beyond consultation toward co-creation. Underrepresented stakeholders should be integrated into working groups, pilot projects, and decision-shaping processes, not only invited to provide input. Broadening participation in these ways will strengthen both the equity and the practical relevance of global AI governance.

To foster meaningful and dynamic engagement, the AI Dialogue should move beyond traditional plenary formats and adopt more interactive, problem-oriented approaches. One promising format is scenario-based simulations. Multi-stakeholder groups could engage in structured exercises, such as responding to AI-enabled misinformation during elections or managing AI-related risks in a natural disaster. These simulations encourage practical collaboration, reveal coordination gaps, and generate actionable insights that purely conceptual discussions often miss. Time-bound policy labs or "sprints" could be introduced. Diverse participants, governments, industry, academia, and civil society, would work intensively over a short period to co-develop specific outputs, such as model guidelines, interoperability frameworks, or crisis response protocols. This format promotes co-creation and tangible deliverables. Regional-to-global dialogue tracks could strengthen inclusivity. Regional workshops held in advance of global sessions would surface context-specific challenges and priorities, which are then integrated into plenary discussions. This helps bridge global norms with local realities. The Dialogue could establish continuous digital collaboration platforms. These would allow stakeholders to share data, best practices, and policy developments in real time, enabling sustained engagement beyond annual meetings. Features such as open repositories, peer review mechanisms, and collaborative drafting tools would enhance transparency and collective learning. Multi-stakeholder review panels could be used to evaluate proposed policies or frameworks from different perspectives, technical, legal, ethical, and societal, ensuring more balanced and robust outcomes. Incorporating youth and practitioner forums, including educators, election officials, and crisis responders, would bring in operational insights often missing from high-level discussions. Together, these formats can transform the Dialogue from a forum for exchange into a platform for joint problem-solving, experimentation, and implementation

Several existing policies and practices offer practical lessons for effective AI governance across safety, capacity-building, and interoperability. Risk-based regulatory frameworks provide a structured approach to managing AI harms. The European Union's EU AI Act is a leading example, introducing tiered obligations based on risk levels, alongside requirements for transparency, human oversight, and accountability. This approach helps prioritize regulatory efforts where risks are greatest while maintaining space for innovation. Multi-stakeholder standards and principles have proven valuable in building shared foundations. The OECD AI Principles and the UNESCO Recommendation on the Ethics of AI have facilitated international convergence on core values, informing both national strategies and corporate practices. Cooperative policy processes such as the G7 Hiroshima AI Process demonstrate how like-minded countries can rapidly coordinate on emerging risks, including generative AI, and develop voluntary codes of conduct that can later inform broader global frameworks. Fourth, operational responses to misinformation and disinformation offer concrete governance models. Practices developed for crisis communication, such as cross-agency coordination, real-time information sharing, and partnerships with platforms, have been increasingly adapted to AI-enabled information risks, particularly during elections and emergencies. Capacity-building initiatives and open ecosystems are critical for inclusivity. Efforts to promote open-source AI models, shared datasets, and technical training programs help lower barriers to participation and enable more countries to engage meaningfully in both deployment and governance. Finally, regulatory sandboxes and pilot projects allow policymakers to test AI applications and governance tools in controlled environments, fostering innovation while managing risks. Taken together, these approaches highlight the importance of combining principle-setting with practical implementation tools, supported by international cooperation and continuous learning.