Trump and Xi agree an AI incident channel but not guardrails

AI was on the agenda when Donald Trump hosted Xi Jinping at the White House, but the two leaders came away with a channel for talking about AI emergencies rather than any shared rules. The groundwork had been laid four days earlier in New York, where US Treasury secretary Scott Bessent and Chinese vice-premier He Lifeng discussed setting up a US–China AI dialogue with a “notification mechanism” for incidents that reach “a national security level”, Euronews reported. “Moving from opaque to more transparency between the No. 1 and the No. 2 AI powers in the world is very important,” Bessent said.

At the summit itself the gap between the two governments’ framing was plain. Xi called for AI to be kept under human control, while Trump rejected demands for new AI rules, a position consistent with his dismissal of risk warnings as a “hoax” ten days earlier. Bessent and He had agreed to meet again to take the dialogue forward. The talks came against a backdrop of sharpening AI rivalry, including a US advisory accusing Chinese labs of distilling American models weeks earlier.

The summit followed pressure from several directions to go further. Senate Democrats led by Maria Cantwell had written to Trump urging him to negotiate formal guardrails on frontier AI with Xi, and a day earlier the heads of OpenAI and Anthropic had urged international standards at the UN Security Council. An incident hotline between the two largest AI powers was nonetheless a first: the kind of crisis-communication channel long argued for by researchers who saw the main risk as runaway systems or misuse crossing borders faster than governments could talk to one another.