Mallaby urges US-China AI pact modeled on nuclear agency

Sebastian Mallaby of the Council on Foreign Relations has proposed establishing an international artificial intelligence regulatory body modeled after the International Atomic Energy Agency to manage US-China strategic competition. Writing for The World Today, Mallaby argues that current geopolitical flashpoints necessitate a framework similar to Cold War-era nuclear nonprotection agreements to prevent catastrophic AI misuse. He contends that both nations face shared existential risks from unregulated advanced systems that transcend commercial rivalry. The proposal emphasizes verifying compliance with safety standards over restricting technological development. This approach seeks to replicate the IAEA’s success in creating inspection regimes and material accounting for dual-use technologies. Mallaby suggests that without such institutionalized cooperation, competitive pressures may incentivize dangerous capability races. The recommendation comes amid escalating export controls and limited diplomatic channels regarding advanced semiconductor access and model weights.