US proposes AI safety notification system in China trade talks

US floats AI incident notification framework

US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng concluded trade discussions in New York on Sunday with an American proposal to establish an AI safety notification mechanism between the world's two largest economies. The framework would be presented to President Donald Trump and Chinese President Xi Jinping for consideration at their summit scheduled for Thursday and Friday in Washington.

According to details first reported by CNN, Bessent told reporters the proposed system would focus on national security concerns, creating a notification protocol for AI-related incidents that reach a threshold of strategic significance. The mechanism aims to increase transparency between what Bessent characterized as "the number one and the number two AI powers in the world."

Why it matters

An AI safety notification system between Washington and Beijing would represent the first formal bilateral channel for managing technology risks that span both nations' interests. With AI development accelerating and both countries racing for technological dominance, the absence of communication protocols creates genuine risks of miscalculation or unintended escalation when incidents occur.

Limited details on Chinese response

The Chinese delegation's position on the AI proposal remains unclear. Vice Premier He and chief trade negotiator Li Chenggang departed the talks at JPMorgan Chase headquarters without addressing media. China's Ministry of Commerce announced Saturday that He would lead a delegation of Chinese companies to the US for economic consultations ahead of the summit.

US Trade Representative Jamieson Greer clarified that existing American export controls on advanced AI chips and semiconductor manufacturing equipment were not part of the AI mechanism discussions. He characterized Sunday's meeting as groundwork for what he hopes will be a "successful summit" later this week.

Broader trade agenda shows incremental progress

Beyond AI, the New York talks addressed implementation of a "Board of Trade" process agreed upon in May, designed to identify non-strategic goods eligible for tariff reductions. Greer indicated potential candidates could include consumer goods and low-tech items from China, while US exports might encompass energy products, agricultural goods, and medical devices.

The discussions did not yield reported progress on several unresolved issues from the May Beijing meeting between Trump and Xi, including Chinese commitments to increase US agricultural purchases by $17 billion annually and orders for more than 200 Boeing aircraft. US officials also did not mention advances on critical mineral supply access, which they have previously described as insufficient despite Beijing's prior commitments.

Congressional pressure on AI safety

US Representative Ro Khanna, ranking Democrat on the House Select Committee on Strategic Competition with China, told CBS on Sunday that Trump and Xi must reach an AI safety agreement. Khanna advocated for international principles including bans on recursive self-improving AI and safeguards preventing AI misuse for biological or nuclear weapons development.

Trump and Xi first discussed potential AI consultations during their May meeting in Beijing, but that forum was never formalized. The current proposal represents a renewed attempt to establish structured dialogue on technology risks between the two powers.

These details were first reported by CNN.