Independent

A successful Global Dialogue on AI Governance would move beyond high-level principles toward greater alignment and coordination across governance regimes. First, progress on interoperability between AI governance approaches would be a meaningful outcome. As different jurisdictions develop their own frameworks, there is a growing risk of fragmentation that increases compliance complexity and slows cross-border innovation. The Dialogue should therefore facilitate structured exchange between regulatory and policy approaches, with the aim of identifying areas of convergence, mutual recognition, and compatibility across regimes. Second, the Dialogue should place stronger emphasis on the institutional and operational conditions required for AI adoption, particularly in emerging markets. In many countries, the primary constraint is not the absence of policy frameworks, but limited implementation capacity, fragmented institutional responsibilities, and insufficient integration of AI into existing public and private sector systems. Without addressing these realities, global governance discussions risk remaining conceptually aligned but operationally hollow. Finally, the Dialogue would be successful if it produces concrete outputs that can be operationalised by governments and institutions e.g., guidance on sequencing policy interventions, institutional coordination models, and approaches to capacity-building. These should be tailored to different levels of digital and economic maturity. Together, these outcomes would help shift global AI governance from a largely normative exercise toward one that supports effective implementation.

AI capacity-building;Interoperability of governance approaches;Transparency, accountability, and human oversight;Social, economic, ethical, cultural, linguistic and technical implications of AI;

My selection reflects on the practical conditions required for effective and scalable AI adoption. AI capacity-building and interoperability of governance approaches are central priorities. Many countries are developing AI strategies, but with varying levels of institutional readiness. Strengthening capacity and ensuring greater alignment across governance frameworks can help reduce fragmentation and support more coordinated global implementation. I have also picked the social and economic implications of AI, as these are critical to understanding how AI systems are integrated into real-world contexts, particularly in emerging markets where adoption pathways differ significantly. Finally, transparency, accountability, and human oversight are essential to ensuring that AI systems are deployable in practice. These elements underpin trust and enable institutions to adopt AI responsibly while maintaining oversight. Together, these priorities reflect the need to move towards operational, context-sensitive approaches to AI governance.

One cross-cutting issue that is not sufficiently captured in the listed themes is the role of cross-border data flows and data governance frameworks in enabling AI development and deployment. AI systems depend fundamentally on access to large, diverse, and often cross-border datasets. However, emerging regulatory approaches, particularly data localisation requirements or restrictions on cross-border data transfers can create significant constraints on model development, testing, and scaling. This is especially relevant in emerging markets, where access to high-quality local datasets may already be limited. Therefore, there is growing tension between national data sovereignty objectives and the need for data openness and cross-border collaboration to support AI innovation. In practice, this can result in fragmented data ecosystems, increased compliance burdens, and uneven access to AI capabilities across jurisdictions. Addressing this issue requires more emphasis on mechanisms for trusted data sharing, interoperability between data protection and AI regulations, and approaches to balancing risk management with the need for innovation. Without greater alignment on cross-border data flows, there is a risk that global AI governance efforts will remain constrained by underlying fragmentation in data governance systems.

I will share my perspective on Asia-Pacific where governance gaps in AI are closely linked to broader disparities in digital maturity and institutional capacity, which shape how countries approach regulation and deployment. A key challenge is the divergence in regulatory approaches, particularly around data governance. Several countries are adopting or considering data localisation measures, while others are pursuing more open or interoperable models. This creates increasing fragmentation across the region, raising costs, limiting collaboration, and constraining access to the diverse datasets required for AI development. At the same time, there are significant gaps in institutional readiness and implementation capacity. While many countries have developed national AI strategies or governance frameworks, translating these into operational systems remains uneven. Challenges include limited technical capacity within public institutions, unclear allocation of responsibilities across agencies, and insufficient integration of AI into existing digital and administrative systems. However, these gaps also present unique opportunities. The diversity of approaches across the region creates space for regional dialogue on interoperability, mutual learning, and the development of more APAC-specific approaches. There is also an opportunity to focus on capacity-building and sequencing of policy interventions, ensuring that governance frameworks are aligned with levels of digital maturity and institutional capability.

The AI Dialogue can play a critical role by shifting the focus of international cooperation from high-level principles toward the institutional and operational conditions required for effective implementation. While many countries are developing AI governance frameworks, there is significantly less clarity on how these can be operationalised in practice, particularly in emerging economies facing constraints in capacity, coordination, and system integration. By facilitating structured exchange on implementation approaches, including sequencing of policies, institutional arrangements, and capacity-building models, the Dialogue can help translate global consensus into actionable pathways. This would enable more meaningful cooperation by ensuring that governance efforts are not only aligned at a conceptual level, but also grounded in the realities of deployment across diverse country contexts.

The AI Dialogue should build on existing multilateral and multi-stakeholder initiatives that are already advancing discussion and collaboration on AI governance, including the ITU's AI for Good initiative, as well as ongoing work within the OECD on AI principles, and the UNESCO Recommendation on the Ethics of Artificial Intelligence. These platforms have contributed to shaping global norms, facilitating dialogue, and sharing best practices across governments, industry, and civil society. The added value of the AI Dialogue would be to connect these existing efforts more systematically and move towards practical coordination. While many initiatives have developed principles and guidance, there remains a gap in how these translate into interoperable governance approaches and operational models across different jurisdictions.

- Structured, multi-stakeholder engagement with defined roles: The Dialogue have targeted working tracks or breakout groups focused on specific issues (e.g., interoperability, capacity-building), with each group contributing concrete inputs and recommendations. This would ensure that diverse perspectives are captured while maintaining focus and accountability. - Outcome-oriented format with iterative outputs: The Dialogue could include short policy briefs, implementation toolkits, or guidance notes developed through iterative consultations before and after the convening. A structured follow-up mechanism would help track progress and maintain continuity between Dialogue sessions.

Communities from emerging and developing regions, particularly across Asia and Africa, remain underrepresented, especially those involved in implementation at national and subnational levels. This includes local institutions, small enterprises, and technical communities working outside major global hubs.

Hybrid formats combining in-person and virtual participation can broaden engagement across regions and time zones. This can be complemented by small-group working sessions with pre-defined outputs and expert moderators to ensure discussions remain focused and action-oriented.

Emerging AI Safety Institutes provide a useful example of institutional approaches to AI governance, focusing on evaluation, testing, and risk assessment of advanced systems. In parallel, frameworks such as the evolving EU AI Act illustrate structured, risk-based approaches to oversight, while sandbox models adopted in several countries enable controlled experimentation. Together, these approaches highlight the importance of combining institutional capacity, regulatory clarity, and iterative learning in advancing effective AI governance while encouraging innovation.