Oxford Martin School AI Governance Initiative, Oxford University

The governance of AI has never been a purely technical matter; it is a geopolitical and ethical challenge that demands coordinated leadership and collective action. The UN High‑Level Advisory Body on AI (HLAB AI) was mandated to propose concrete institutional and strategic pathways for global oversight. Its Final Report, Governing AI for Humanity, set out seven core recommendations aimed at fostering inclusive, rights‑respecting, effective global governance. The seven recommendations spanned two themes. Theme 1 considers the institutional foundations needed for credible global governance, while Theme 2 examines the operational infrastructure and equity mechanisms required to translate principles into practice. Although implementation efforts are underway, including the Scientific Panel and Global Dialogue, progress remains uneven, and reliance on voluntary measures continues to limit enforceability and coherence. To recapture the ambition of the HLAB AI mandate, the Global Dialogue could, and I would argue should, advance proposals for a renewed approach and identifies where emerging consensus places leadership for each strand of work. For all seven recommendations, the activity could set out which intergovernmental body or multistakeholder group is currently positioned to drive implementation. The aim is to provoke buy-in and strengthen ownership, clarify responsibility, and support more coordinated momentum.

• AI capacity-building
• Interoperability of governance approaches

The HLAB-AI called for a distributed infrastructure to help countries build the skills, resources, and institutional capabilities needed to govern and benefit from AI and thereby close the AI capacity gap. Progress has come through the ITU's AI for Good initiative and UNITAR / ITU Academy training, offering modules on AI ethics, governance, and technical skills for public servants and social entrepreneurs. The EU has also signalled support for digital commons and public infrastructure via its Global Gateway and Next Generation Internet programs. Technology companies have also contributed, with Google AI Opportunities Fund and Microsoft's AI for Good Lab supporting capacity-building aligned with the UN SDGs. Despite these developments, the vision of a globally interconnected capacity-building infrastructure remains nascent. UNESCO's own AI Ethics Observatory highlights disparities in national readiness, underscoring the need for a coherent and scalable framework

Since the UN HLAB-AI report's release, the ITU launched the AI Standards Exchange Database. Conceived as a collaborative space for standards bodies, tech firms and civil society, it now serves mainly as an exhaustive catalogue of existing standards. It lacks authority to resolve conflicts or coordinate across jurisdictions. Reliance on voluntary input undermines accountability, while weak integration with emerging models, impact assessments, and sandboxing limits its relevance for forward-looking policy and engagement with Big Tech. To fulfil its promise, critics propose that the database exchange must evolve beyond aggregating standards to mediating between them and anchoring accountability in binding frameworks. Its short-term success depends on voluntary engagement; its long-term viability rests on evolving into a dynamic, inclusive, and authoritative platform.

I do not see the topic thru a country, region or sector lens. These are all important but nothing will work unless an International Dimension is also established. This returns to the original mandate of the UN HLAB-AI: The governance of AI is not merely a technological concern; it is a geopolitical and ethical challenge that demands focused leadership & effective collective action. Recognizing this, in 2024 the UN Secretary-General António Guterres established a High-Level Advisory Body on Artificial Intelligence (HLAB-AI) entrusted, through its Terms of Reference with a bold and consequential mission: to deliver actionable recommendations on the core functions, institutional design, and strategic timeline for launching a dedicated global agency to oversee AI governance.

To raise the profile of the necessary International aspect of AI Governance / Regulation. All layers are needed but only when they work together - with Big Tech in the room - will meaningful progress be made. The governance of AI is not merely a technological concern; it is a geopolitical and ethical challenge that demands focused leadership & effective collective action. Recognizing this, in 2024 the UN Secretary-General António Guterres established a High-Level Advisory Body on Artificial Intelligence (HLAB-AI) entrusted, through its Terms of Reference with a bold and consequential mission: to deliver actionable recommendations on the core functions, institutional design, and strategic timeline for launching a dedicated global agency to oversee AI governance.

Return to the excellent work carried out by the original UN HLAB-AI Final Report. 27 global experts got the report thru the UN Summit of the Future ... this work should be used as the template for what now needs to happen - step-by-step in the implementation phase. Major structural tensions continue to impede progress: geopolitical fragmentation, institutional inertia, and reliance on voluntary mechanisms. Strategic competition between the United States and China further obstructs convergence. Meaningful governance, as envisioned by the HLAB AI, cannot emerge from such a landscape. In this environment strong commercial actors wield disproportionate influence. What is missing is not vision or launch events, but durable institutional design. The time for cautious optimism has passed; what is required now is strategic diplomacy, sustained engagement, and the resolve to move beyond rhetoric into enforceable practice.

Use the seven recommendations (of which the global dialogue itself was one) to structure the intended dialogue outcomes

• Like the establishment of the IAEA, CERN, and ITER, the realization of the IAIA (or any global forum for AI Governance / Regulation oversight) will require a blend of diverse skillsets. Identifying and inspiring individuals with the necessary interpersonal skills to build mutual trust will be crucial. We need to consider three groups: first and most importantly, political leaders who need to provide the global vision and be enablers and champions
• secondly, world-renowned technological and academic experts who need to provide credible ideas to make the body work
• and finally accomplished diplomats and administrators who can make it happen.

• Modern diplomats navigate complex international relations, represent their country's interests, and help maintain peaceful cooperation between states
• they also pride themselves on getting things done. History has recorded the herculean efforts of those who helped create the IAEA , CERN and ITER
• it was high-level strategic lobbying by like-minded individuals who had personal relationship developed over decades that held the key to IGO approaches gaining traction over alternatives. There is an edge of competitiveness and constructive rivalry between the existing AI governance set-ups that may be able to be harnessed. The diplomats who play a key role in the establishment of an IAIA will develop reputations that will be the envy of their colleagues.

Establishing an agency to apply international governance to a potentially global existential threat has been done before. The International Atomic Energy Agency (IAEA) was formed at the height of tensions between its two principal members of the period: the USA and USSR. Since then, the Agency has earned a reputation for impartiality which all the Member States closely monitor but it also continues to face operational challenges. Kissinger and Allison while urging world leaders to "act now as squarely as their predecessors addressed nuclear threats in earlier decades", also charted several practical difficulties with regulating AI in comparison to the nuclear domain. This input recognises those shortfalls and does not advocate the IAEA as a perfect model. However, we firmly make the case that, such is the importance and reach of AI, the underlying principle that its governance must fall under the auspices of the UN is increasingly irrefutable. Achieving a sustainable, inclusive, and fair international regulation of the 'Coming Wave' of AI is one of the most important questions of our times. The way ahead to achieve that is emerging through collaborative initiatives such as that undertaken by the UN AIAB. This paper contributes to that effort by being bold in outlining one way forward to actively engage the three groups that would need to work together to make the Agency a reality. A successful global AI regulatory framework regime, based on structures of international collaboration that have been outlined here could herald an era of global cooperation that could finally enact the same high ideals envisaged by the founders of the UN; it is that important to get it right. All of this will make the achievement truly memorable and meaningful.