AI & Partners

A successful first Global Dialogue on AI Governance would achieve several practical and strategic outcomes. First, it should establish a shared baseline of principles for responsible AI development and use. While countries differ in priorities, agreement on core values—such as transparency, accountability, safety, human rights protection, and fairness—would create a common foundation for future policy alignment. Even a non-binding declaration could help guide national regulations and corporate practices. Second, the dialogue should produce clear mechanisms for ongoing international cooperation. AI development is global, so governance cannot remain fragmented. A successful outcome would include commitments to continued multilateral dialogue, working groups on key issues (such as safety standards, data governance, and frontier AI risks), and structured channels for information sharing between governments, researchers, and industry. Third, it should prioritize inclusion of diverse perspectives, particularly from developing countries. Many nations risk being affected by AI systems without having a voice in how they are governed. Ensuring meaningful participation from the Global South, civil society, academia, and the private sector would strengthen legitimacy and produce more balanced governance approaches. Fourth, the dialogue should move beyond principles toward practical cooperation, such as collaboration on AI safety research, capacity-building programs, and shared technical standards. Early cooperation in areas like risk assessment, model evaluation, and incident reporting could build trust among stakeholders. Finally, success would be measured by whether the event builds momentum rather than ending as a one-time discussion. Establishing a roadmap for future meetings, policy coordination, and measurable follow-up actions would ensure the dialogue becomes a lasting platform for global AI governance. In essence, the first Global Dialogue would be successful if it lays the groundwork for coordinated, inclusive, and forward-looking international governance of AI, while fostering trust among governments, industry, and society.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Interoperability of governance approaches

From my perspective, the four thematic areas that require the most urgent action and active engagement are Safe, Secure and Trustworthy AI, AI Capacity-Building, Transparency, Accountability and Human Oversight, and Interoperability of Governance Approaches. Safe, Secure and Trustworthy AI is the most immediate priority because AI systems are already being deployed across critical sectors such as healthcare, finance, education, and public administration. Without robust safety standards and risk management frameworks, AI systems can cause unintended harm, amplify bias, or be misused. Urgent action is needed to develop shared safety benchmarks, evaluation methods for advanced models, and mechanisms for monitoring and mitigating risks. AI Capacity-Building is equally essential to ensure that all countries-not only technologically advanced ones-can participate meaningfully in the AI ecosystem. Many developing countries face gaps in infrastructure, expertise, and institutional readiness. Supporting education, technical training, research collaboration, and digital infrastructure will help reduce global inequalities and enable responsible AI adoption that supports sustainable development goals. Transparency, Accountability and Human Oversight are critical for building public trust in AI systems. People affected by AI-driven decisions should have access to clear explanations of how those decisions are made and the ability to challenge harmful outcomes. Strengthening transparency requirements, accountability mechanisms, and ensuring meaningful human oversight can help protect rights and ensure AI systems remain aligned with societal values. Finally, Interoperability of Governance Approaches is vital because AI development and deployment operate across borders. Fragmented regulatory frameworks can create uncertainty, limit collaboration, and reduce the effectiveness of governance efforts. Promoting interoperability between national and regional AI regulations can help harmonize standards, facilitate innovation, and enable coordinated responses to emerging risks. Together, these four priorities support a balanced approach to AI governance-one that promotes innovation while safeguarding human rights, global equity, and long-term societal well-being.

Yes. While the listed themes capture many core aspects of AI governance, several cross-cutting and emerging issues also deserve greater attention. Environmental sustainability of AI is an increasingly important issue. Training and operating large AI models can require significant energy and water resources, contributing to carbon emissions and environmental strain. As AI adoption accelerates, governance frameworks should encourage energy-efficient model design, transparent reporting of environmental impacts, and the use of sustainable infrastructure. Another key issue is data governance and data equity. AI systems rely heavily on large datasets, yet access to high-quality data is uneven across regions and sectors. Questions around data ownership, consent, privacy, and fair compensation for data contributors remain unresolved. Strengthening global cooperation on responsible data sharing and equitable access will be essential to ensure that AI development benefits a wider range of societies. Economic and labor market impacts also represent a critical cross-cutting challenge. AI has the potential to transform industries and reshape employment patterns, creating both opportunities and disruptions. Governance discussions should therefore include strategies for workforce reskilling, social protection, and policies that help societies adapt to AI-driven economic change. Finally, misinformation and the integrity of information ecosystems are rapidly emerging concerns. Generative AI systems can produce highly realistic synthetic media and automated content at scale, which may undermine trust in public information, democratic processes, and media institutions. Addressing these risks requires collaboration among governments, technology companies, civil society, and researchers to develop safeguards, detection tools, and responsible use standards. Addressing these cross-cutting issues alongside the existing themes would strengthen global AI governance by ensuring that discussions consider not only technical and regulatory challenges but also the broader societal, economic, and environmental implications of AI systems.

Governance gaps in the priority areas of safe and trustworthy AI, capacity-building, transparency and accountability, and interoperability of governance approaches are already shaping both challenges and opportunities across many sectors. One major challenge is the uneven development of AI governance frameworks across countries and regions. Different regulatory approaches and standards can create uncertainty for organizations developing or deploying AI systems, particularly those operating internationally. This fragmentation can slow innovation, complicate compliance, and make it difficult to ensure consistent safeguards for users. Another key challenge relates to limited technical capacity and institutional readiness. Many governments, institutions, and organizations lack sufficient expertise, infrastructure, and regulatory tools to effectively oversee rapidly evolving AI technologies. This gap can make it harder to evaluate AI risks, implement oversight mechanisms, or support responsible adoption across sectors such as healthcare, education, and public administration. Concerns around transparency and accountability also remain significant. As AI systems become more complex and integrated into decision-making processes, it can be difficult for users, regulators, and affected communities to understand how these systems operate or challenge harmful outcomes. This can undermine public trust and slow the responsible deployment of beneficial AI applications. At the same time, these governance gaps present important opportunities. There is growing momentum for international cooperation on AI safety standards, risk management practices, and shared technical benchmarks. Strengthening collaboration among governments, academia, civil society, and industry can accelerate the development of practical governance tools and best practices. In addition, expanding AI capacity-building initiatives—including education, research partnerships, and technical training—can empower more countries and sectors to participate in the global AI ecosystem. With the right governance frameworks in place, AI also offers opportunities to improve public services, support innovation, and address complex global challenges, from healthcare delivery to climate analysis.

The AI Dialogue can play a critical role as a neutral and inclusive platform for international coordination on AI governance. Because artificial intelligence is developed and deployed across borders, effective governance requires sustained collaboration between governments, international organizations, the private sector, academia, and civil society. First, the AI Dialogue can facilitate consensus-building on shared principles and standards for responsible AI. By bringing together diverse stakeholders, the Dialogue can help align national approaches around core values such as safety, human rights, transparency, and accountability, while respecting different legal and cultural contexts. Second, the Dialogue can support knowledge exchange and policy learning. Countries and organizations are experimenting with different regulatory models, technical standards, and oversight mechanisms. The AI Dialogue can serve as a platform to share lessons learned, identify best practices, and promote evidence-based policymaking. Third, it can strengthen international cooperation on AI safety and risk management. This may include collaboration on model evaluation methods, incident reporting mechanisms, and research on frontier AI risks. Coordinated approaches can help reduce duplication of efforts and ensure that emerging risks are addressed collectively. Fourth, the AI Dialogue can advance inclusivity and capacity-building, particularly for developing countries. Ensuring that all regions have opportunities to participate meaningfully in AI governance discussions will improve the legitimacy and effectiveness of global governance frameworks. Finally, the Dialogue can act as a bridge between policy discussions and practical implementation, helping translate high-level principles into concrete actions such as joint initiatives, technical working groups, and voluntary cooperation frameworks.

The AI Dialogue should build upon and connect with a range of existing international initiatives, policy frameworks, and multi-stakeholder partnerships that are already advancing discussions on AI governance. One important reference point is the work of the Organisation for Economic Co-operation and Development (OECD) on the OECD AI Principles, which have influenced many national AI strategies and governance frameworks. Similarly, initiatives such as the Global Partnership on Artificial Intelligence (GPAI) bring together governments and experts to advance responsible AI development and research collaboration. Regional regulatory efforts also provide valuable lessons. For example, the European Union's AI Act represents one of the most comprehensive attempts to regulate AI based on risk categories. Other national and regional strategies around the world are experimenting with regulatory sandboxes, safety standards, and oversight institutions. Technical and industry-led initiatives are also important. Organizations such as the International Organization for Standardization (ISO) and the International Telecommunication Union (ITU) are working on technical standards that support safe and interoperable AI systems. In addition, research collaborations and safety-focused initiatives across academia and industry contribute to developing evaluation methods, benchmarks, and governance tools. The AI Dialogue can add value by connecting these fragmented efforts within a broader global governance framework. Rather than duplicating existing initiatives, it can serve as a coordinating platform that encourages policy coherence, identifies gaps, and promotes interoperability between different governance approaches. Furthermore, the AI Dialogue can ensure stronger inclusion of developing countries and underrepresented stakeholders, many of whom are not fully represented in existing initiatives. By linking technical, policy, and development-focused discussions, the Dialogue can help create a more balanced and globally representative ecosystem for AI governance.

Different stakeholders can contribute to the AI Dialogue by bringing their unique expertise, experiences, and perspectives. Governments can share regulatory approaches, national AI strategies, and policy lessons, helping identify areas for international alignment. Industry actors provide insights into practical deployment challenges, safety testing, and technical innovation, ensuring discussions remain grounded in operational realities. Academia and research institutions contribute evidence-based analysis, risk assessment frameworks, and cutting-edge technical knowledge. Civil society organizations can highlight societal concerns, ethical implications, and human rights perspectives, ensuring that AI governance remains inclusive and socially responsible. International organizations can facilitate coordination across regions, offer normative guidance, and support capacity-building initiatives. For the format and structure, the Dialogue should be multi-layered and interactive. A combination of plenary sessions for broad principle-setting and thematic working groups for deep dives into topics such as AI safety, interoperability, or transparency would allow both strategic discussions and technical problem-solving. Roundtables and scenario-based exercises could enable participants to collaboratively address emerging challenges, simulate policy impacts, and explore practical solutions. It should also include mechanisms for continuous engagement beyond a single event, such as digital platforms for knowledge-sharing, intersessional working groups, and progress reporting. A rotating co-chair or facilitation model could ensure inclusivity, giving different regions and stakeholder groups the opportunity to shape agendas. Finally, structured opportunities for multi-stakeholder dialogue, with clear pathways for recommendations to influence policy and practice, would maximize the Dialogue's impact, ensuring it is not merely consultative but actionable.

Several groups remain underrepresented in global AI governance discussions. Developing countries and the Global South often lack the technical capacity, resources, or institutional mechanisms to participate fully, despite being directly affected by AI deployment. Small and medium-sized enterprises (SMEs), which are crucial innovators but may not have dedicated policy teams, are rarely represented. Civil society organizations and grassroots movements, particularly those focused on human rights, social equity, and marginalized communities, often have limited access to formal international fora. Indigenous communities, local knowledge holders, and vulnerable populations are also frequently excluded, though AI can disproportionately affect them through biased systems or algorithmic decision-making. To include these voices, the Dialogue should adopt proactive outreach and facilitation measures. Providing financial and technical support for participation from underrepresented countries or organizations can reduce barriers. Creating regional hubs and preparatory consultations can allow stakeholders to consolidate input before global sessions. Digital platforms enabling asynchronous participation, feedback, and collaborative drafting can expand reach beyond those who can travel. Structured formats, such as inclusive advisory committees or observer roles, ensure marginalized perspectives influence agenda-setting and outcomes. Capacity-building initiatives linked to the Dialogue, including training, mentorship, and knowledge-sharing networks, can further empower underrepresented groups to contribute meaningfully. By designing both structural inclusivity and practical support mechanisms, the Dialogue can ensure that global AI governance reflects a broader, more equitable range of perspectives.

To foster meaningful and dynamic engagement, the AI Dialogue could combine interactive, multi-modal formats that go beyond traditional plenary sessions. Scenario-based workshops allow participants to simulate AI deployment challenges, ethical dilemmas, or regulatory responses, encouraging practical problem-solving and collaborative learning. Hackathons or design sprints could bring technical experts, policymakers, and civil society together to co-create prototypes, tools, or governance frameworks in real time. Thematic roundtables with rotating moderators can ensure focused discussions on high-priority areas, such as AI safety, transparency, or cross-border governance, while allowing diverse stakeholders to contribute without being overshadowed by larger delegations. Digital participation platforms with live polls, Q&A, and collaborative document editing can extend engagement to stakeholders who cannot attend in person, creating an ongoing participatory ecosystem. Another innovative approach is "citizen deliberation sessions", where representatives of affected communities or the public discuss AI impacts and provide feedback to policymakers. These can be integrated with expert panels to ensure societal concerns are considered alongside technical and regulatory perspectives. Finally, establishing inter-sessional working groups or thematic labs enables continuity, allowing insights from the Dialogue to evolve into actionable recommendations over time. Visual tools, such as interactive dashboards and policy mapping exercises, can help participants track discussions, compare governance approaches, and identify opportunities for alignment. Combining these formats—hands-on exercises, digital collaboration, inclusive deliberation, and continuous engagement—would create a dynamic, inclusive, and action-oriented Dialogue, ensuring that participants not only share ideas but also co-develop practical solutions for international AI governance.

Several policies, practices, platforms, and approaches around the world demonstrate effective AI governance and offer practical solutions to emerging challenges. 1. Risk-based regulatory frameworks: The European Union's AI Act establishes a tiered risk approach, categorizing AI systems by potential harm and imposing requirements accordingly. This approach balances innovation with safety, ensuring that high-risk AI systems-such as those used in healthcare, law enforcement, or critical infrastructure-are subject to stricter oversight. 2. Multi-stakeholder initiatives: Platforms like the Global Partnership on Artificial Intelligence (GPAI) and OECD AI Principles foster collaboration between governments, industry, and academia. These initiatives provide guidance on responsible AI development, facilitate knowledge exchange, and promote harmonized approaches across borders. 3. Technical standards and auditing practices: Organizations such as ISO and IEEE have developed technical standards for AI systems, including guidelines on transparency, robustness, and bias mitigation. Complementary practices, like independent algorithmic audits and model documentation (e.g., "Model Cards"), enhance accountability and allow stakeholders to assess AI performance, fairness, and risks. 4. Regulatory sandboxes: Countries like the United Kingdom and Singapore have implemented AI regulatory sandboxes that allow developers and regulators to test AI systems in controlled environments. This approach encourages innovation while monitoring safety and compliance, generating lessons for future regulation. 5. Transparency and human oversight mechanisms: Practices such as mandatory explainability reports, impact assessments, and human-in-the-loop requirements ensure that AI systems remain understandable, controllable, and aligned with ethical and societal standards. 6. Capacity-building and inclusive governance: Programs such as AI4D (Artificial Intelligence for Development) support research, technical training, and policy capacity in underrepresented regions, promoting equitable access to AI knowledge and enabling broader participation in governance discussions. Collectively, these policies and approaches demonstrate that effective AI governance requires a combination of regulation, technical standards, stakeholder collaboration, and capacity-building, all designed to ensure that AI systems are safe, transparent, accountable, and socially beneficial.