International Institutions for Advanced AI

International coordination can reduce risks while spreading benefits: The paper argues that advanced AI systems create "global externalities" in their development and deployment, meaning risks and benefits cross borders. International institutions could help manage dangerous capabilities while enabling AI to support sustainable development goals.

Four complementary institutional models are proposed: The paper identifies specific governance structures with real-world precedents: a Commission on Frontier AI for expert consensus, an Advanced AI Governance Organization to set safety standards, a Frontier AI Collaborative to ensure equitable access, and an AI Safety Project to advance safety research.

Safety standards need international agreement: Rather than countries acting alone, the paper suggests an international body could establish common safety standards for advanced AI systems, reducing fragmentation and ensuring responsible practices are implemented consistently across borders.

Access to AI technology matters for equity: One proposed institution—the Frontier AI Collaborative—specifically aims to promote access to cutting-edge AI systems, suggesting the paper recognizes that concentrating advanced AI in few hands could exacerbate global inequalities and limit AI's potential to address worldwide challenges.

Existing international organizations provide a roadmap: By grounding these proposals in precedents from established international institutions, the paper suggests these governance models are feasible rather than purely theoretical, offering practical templates for how countries might cooperate on AI governance.