US and Chinese experts propose nuclear-style AI safeguards, including a dedicated AI military hotline, as part of a dialogue ahead of planned bilateral talks

Accidental AI-driven escalation in military operations may soon be preventable through direct U.S.-China communication channels. If a dedicated hotline for AI incidents is established, it could de-escalate potential conflicts that arise from autonomous system misinterpretations.

U.S. and Chinese security experts have proposed a set of safeguards for military artificial intelligence (AI) to prevent unintended escalation, particularly concerning nuclear command systems and critical infrastructure. Recommendations include maintaining human control over AI-enabled cyberattacks targeting nuclear command or vital national systems, and establishing a dedicated U.S.-China military hotline for incidents involving autonomous AI. These proposals emerge ahead of planned bilateral government talks on AI and a potential meeting between U.S. President Donald Trump and Chinese President Xi Jinping. The suggestions are detailed in work by Melanie Sisson of the Brookings Institution and Tianjiao Jiang of Fudan University, who advocate for restrictions on AI operating near nuclear systems and a shared definition of "meaningful human control." Both nations are cautiously assessing these ideas, balancing risk mitigation with the desire to maintain technological advantages in AI development.

The proposals signal a significant shift in the U.S.-China AI rivalry, moving beyond competition over hardware and models to address existential risks akin to nuclear arms control. Experts are now prioritizing preventing accidental escalation, machine-driven military decisions, and ensuring clear communication during crises, reflecting a growing recognition of AI's destabilizing potential in military contexts. This focus on risk management, especially concerning nuclear systems, suggests a new paradigm in how great powers are beginning to conceptualize the governance of advanced military technologies.

The push for a dedicated AI hotline and shared definitions of human control highlights the urgent need for clear protocols in a domain where decisions can outpace human reaction times. This initiative impacts all nations developing or operating AI in security sectors, as it sets a precedent for international cooperation on managing emergent technological risks. Future developments will likely focus on the practical implementation of these safeguards and whether existing communication channels can be adapted or if entirely new ones are required.