The United States is proposing a direct notification system with China for artificial-intelligence incidents that affect national security. The concept resembles the Cold War hotline with Moscow: either government could quickly alert the other when an AI event creates a serious risk of misunderstanding or escalation.
US officials expect negotiations to take weeks and hope to have a working framework approved before Treasury Secretary Scott Bessent meets Chinese Vice Premier He Lifeng in Shenzhen in roughly two months. China has remained noncommittal, and the two sides have not defined what kind of model behavior, cyber incident or military concern would require notification.
Enforcement is another unresolved issue. American officials want obligations that apply to both countries rather than a channel used mainly for US warnings. The talks are also entangled with trade and chip policy, including limited US licenses for Nvidia H200 sales to China and debate over whether export restrictions accelerate domestic Chinese alternatives.
A hotline would provide communication, not an agreement on AI development or safety standards. It could reduce the danger of one government misreading an incident, but only if both sides establish trusted contacts, meaningful thresholds and a willingness to disclose sensitive events. For now, the proposal is still a diplomatic framework rather than an operational safeguard.