US Treasury Secretary Scott Bessent said after meeting Chinese Vice Premier He Lifeng in New York on September 20 that the world's two leading AI powers discussed creating a standing channel called the "US-China AI dialogue." Washington proposed a mechanism for the two countries to notify each other of AI "incidents" that could escalate to a national-security level, and both sides agreed to meet again. The talks were a prelude to the September 24 White House meeting between President Donald Trump and President Xi Jinping — a rare signal of cooperation amid an intensifying race for AI supremacy.

The United States and China have discussed establishing a permanent channel to talk about artificial intelligence. Speaking to reporters after talks with Chinese Vice Premier He Lifeng in New York on September 20, US Treasury Secretary Scott Bessent said the two sides addressed setting up a way to communicate about AI issues, referred to as the "US-China AI dialogue." The meeting served as groundwork for the Trump-Xi summit scheduled at the White House on the 24th.

Washington's 'Incident Notification' Proposal

"The US has proposed that we have a notification mechanism between the two countries," Bessent said. Such notifications, he explained, would cover potential "incidents" — situations "that rise up to a national security level from AI." He added that both sides want "a shared vision of common goals and common threats," and stressed that moving the relationship between the world's number one and number two AI powers "from opaque to more transparency" is very important. The two sides agreed to meet again on the matter.

Date · Location September 20, 2026 (Sun) · New York, USA
US side Treasury Secretary Scott Bessent (with USTR Jamieson Greer)
China side Vice Premier He Lifeng
Core proposal Mutual notification of national-security-level AI "incidents"
Next steps Both agree to meet again · to be raised at the Sept 24 Trump-Xi White House meeting

Why Rivals Are Cooperating Now

The discussion comes as fears grow in both countries that increasingly capable AI systems could pose existential risks — a rare incentive for cooperation even as the contest for technological supremacy intensifies. Trump, however, has resisted calls to slow AI development, arguing that doing so would give China room to catch up with US firms. That logic frames the proposal on the table: keep competing, but tell each other when something goes wrong.

China's Response and the Open Questions

China's state news agency Xinhua, in a brief readout, said the two sides "engaged in discussions on key economic and trade issues of mutual interest, such as implementing the consensus from previous consultations and promoting bilateral trade and investment." Xinhua added that officials also held a dialogue on "issues related to AI," but gave no further details. Whether the notification mechanism becomes an actual arrangement now hinges on how Trump and Xi receive it at the September 24 summit.

Item Detail
Status A "proposal" for a channel and notification mechanism — not a binding agreement
Scope Limited to AI "incidents" that rise to a national-security level
US position Keep the pace of competition; expand transparency on safety
China position Confirms "discussions on AI-related issues," no specifics disclosed
Next step To be raised at the Sept 24 Trump-Xi White House meeting · both agree to meet again
The notable point is that this is not "AI arms control" or a development cap — it resembles a crisis hotline, closer to how nuclear or aviation incidents are reported between states. It reads as a Cold War-era crisis-management model applied to AI: don't stop competing, but prevent catastrophe from miscalculation.

What It Means

The mere fact that the US and China have formalized AI as a summit agenda item and discussed a standing communication channel is unusual. Until now, their AI touchpoints have centered on export controls (chips and equipment) and mutual regulation; this is the first time a cooperative frame — "safety incident notification" — has surfaced. It is, of course, only a proposal, with hurdles ahead from defining what counts as a "national-security-level incident" to verification and enforcement. Even so, by leaving room for global AI norms not to collapse entirely into bloc confrontation, the outcome of the September 24 summit will be an early test of where international AI governance is headed.

For middle-power AI nations, including South Korea, this trend is far from someone else's problem. If the US and China set bilateral norms like "incident notification" first, the draft of international standards may be settled between the two giants, with everyone else left to follow.
Related Reading · Official Sources
· Euronews — US and China to seek 'AI dialogue' to communicate shared concerns (via AP)
· CNN Business — Bessent proposes AI safety notifications in talks with China ahead of Xi-Trump meeting
· The Detroit News — Bessent proposes US-China AI safety notifications in talks with Chinese vice premier
  • Treasury Secretary Bessent revealed a discussion to create a "US-China AI dialogue" after meeting Vice Premier He Lifeng in New York on Sept 20
  • The US proposed a mechanism to mutually notify each other of national-security-level AI "incidents" — both sides agreed to meet again
  • The talks were groundwork ahead of the Sept 24 Trump-Xi White House meeting
  • "Keep competing, expand safety transparency" — closer to a crisis hotline than a cap on development speed
  • Still only a proposal, with defining, verifying and enforcing "incidents" unresolved; the summit outcome will shape international AI governance