Kurt Campbell says US-China AI talks won’t yield binding limits
Former deputy secretary of state cites deep mutual distrust as the core obstacle to enforceable AI agreements between Washington and Beijing
Former US Deputy Secretary of State Kurt Campbell has warned that discussions between Washington and Beijing on AI and security are unlikely to produce binding agreements, pointing to a fundamental trust deficit that no amount of diplomacy has managed to close.
Notification, not regulation
Rather than pursuing formal restrictions, recent discussions have centered on risk notification mechanisms. Treasury Secretary Scott Bessent led discussions in New York on September 20, proposing the creation of a “US-China A.I. dialogue.” The framework would establish a system where each country notifies the other about national security-level AI incidents. It’s transparency-oriented, not regulatory. Notably, the proposal steered clear of export controls on advanced semiconductors.
Prior dialogues included 2024 talks in Geneva focused on nuclear command risks, and a communication channel established after the Trump-Xi summit in Beijing in May 2026.
Campbell’s core argument is straightforward: neither side trusts the other enough to agree to restrictions they can’t independently verify.
Who’s at the table
The discussions have drawn participation from major US AI firms including OpenAI and Nvidia, alongside senior officials from both governments.
The news moving money, markets, and the world—before your day starts.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Nvidia’s presence is particularly significant. The chipmaker sits at the center of the AI supply chain, and its high-end GPUs have been subject to US export restrictions targeting China since 2022. The fact that semiconductor export controls weren’t part of the proposed dialogue framework suggests both sides are keeping the most contentious issues off the table.
The trust problem runs deep
Washington generally favors a model where private companies lead development under light regulatory oversight, with targeted export controls to limit adversary access. Beijing prefers state-directed AI development with fewer constraints on government surveillance applications but tighter control over private sector autonomy.
The Geneva talks in 2024 on nuclear command-and-control risks offer a useful comparison. Those discussions addressed a domain where both sides had decades of arms control experience and well-understood verification methods.
What to watch
The semiconductor supply chain remains the most potent pressure point. Export controls on advanced chips have done more to shape China’s AI trajectory than any diplomatic conversation. If the notification framework explicitly excludes chip restrictions, it signals that Washington intends to keep that lever separate, available for unilateral use rather than subject to bilateral negotiation.