US-China draft AI nuclear rules aim to prevent autonomous weapon decisions

US and China experts proposed shared rules to ban AI autonomy in nuclear weapon systems, building on a November 2024 agreement but facing practical implementation gaps.

US and China experts Melanie Sisson of Brookings Institution and Tianjiao Jiang of Fudan University proposed shared rules to prevent AI systems from making autonomous decisions regarding nuclear weapon deployment. The rules require humans to retain sole control over AI-driven cyberattacks targeting strategic infrastructure and explicitly prohibit AI from autonomously launching nuclear weapons or attacking nuclear command systems. Both nations must also agree on a shared definition of 'human control.' These proposals build on a US-China agreement reached in November 2024, with Jiang suggesting an AI incident hotline to clarify accidental system responses before escalation. However, the 2023 spy balloon crisis demonstrated the hotline’s limitations when China did not respond to a US call, and Johns Hopkins University’s Carla Freeman questioned the practical effectiveness of such hotlines.

The mechanism centers on establishing mutual technical standards for human oversight in critical systems, aiming to eliminate autonomous escalation pathways during nuclear crises. This directly addresses the risk of AI-driven miscommunication triggering catastrophic conflict.

By preventing AI autonomy in nuclear decision-making, these rules reduce the potential for catastrophic resource destruction from nuclear war—a primary threat to global food security, clean water access, and infrastructure stability. The absence of autonomous systems could preserve critical resources for survival needs during emergencies.

What to watch: Implementation barriers, including the 2023 incident’s demonstration of hotline limitations and the lack of binding mechanisms to enforce shared definitions of 'human control.' The rules remain in development, not yet operational.

Source: The Decoder