AI development is global but governance is national, and the incentives to race are strong. Proposals range from an IAEA-for-AI to compute monitoring treaties, each with obvious enforcement problems. Is there any international governance model you think could actually work, or is coordination fundamentally hopeless?