Restructuring International Organizations

The HLAB-AI final report proposes seven institutional mechanisms, several with direct economic implications: a Global Fund on AI for the SDGs to channel resources toward development goals; an AI Capacity Development Network to build technical expertise in underserved regions; and a Global Data Framework addressing the economic value of data across borders. The report emphasizes a "distributed architecture" rather than a single centralized body, acknowledging that AI governance requires coordination across multiple functional areas—including labor, trade, taxation, and development—that no single institution can address. Notably, the report frames AI governance partly as a development challenge, warning that without intervention, AI could widen rather than narrow global inequality.

The Carnegie Endowment argues that "rather than a single, tidy, institutional solution to govern AI, the world will likely see the emergence of something less elegant: a regime complex, comprising multiple institutions within and across several functional areas." For economic governance, this implies coordination among existing bodies — the ILO for labor standards, the WTO for trade, the OECD for taxation, the IMF and World Bank for macroeconomic stability and development — rather than creating a new comprehensive AI economic authority.

Robinson proposes the establishment of an International Artificial Intelligence Agency (IAIA) under UN auspices, with a governance structure adapted from the IAEA but critically modified to include big tech corporations as "associate members" alongside member states. His proposed 35-member governing board would comprise 20 state representatives and 15 big tech representatives, with no veto powers, creating what he terms a "new Grand Bargain" for AI analogous to the NPT's balance between nuclear and non-nuclear states. Robinson argues that only a UN-convened body can legitimately oversee global AI governance involving all major powers, and that integrating big tech into formal governance structures, rather than the current transient, consultative arrangements, is necessary for credibility and efficacy.

This analysis argues against creating new centralized AI institutions, contending that strengthening the existing "weak regime complex" of international bodies is the more politically legitimate and viable path forward. The authors identify two categories of barriers to global AI governance: first-order cooperation problems stemming from interstate competition (particularly the US-China rivalry over AI's perceived strategic importance) and second-order cooperation problems from dysfunctional international institutions. They recommend foregrounding the OECD as a center of expert AI knowledge to facilitate peer pressure and policy harmonization among states, while scrutinizing whether existing institutions are fulfilling appropriate functions given their democratic mandates.