What happened
During Xi Jinping's state visit to Washington (23-25 September 2026), the White House confirmed the two countries agreed to establish a formal 'bilateral communication channel' for AI incidents, operationalising the US-China AI Dialogue first agreed at officials level on 20 September with a mechanism for notifying AI incidents rising to the national-security level. US Trade Representative Greer compared the channel to a Cold War 'red phone' and said another AI meeting is expected within a month; leaders are due to meet again at APEC (Shenzhen) and G20 (Miami).
Why it matters
This is the first binding heads-of-government AI incident-notification mechanism between the world's two leading AI powers. It creates a concrete cross-border expectation that serious AI safety incidents (e.g. agent containment failures) be disclosed bilaterally, and sets the diplomatic framework within which US and Chinese AI governance and export-control positions will be negotiated.
Action needed
Frontier AI labs and AI infra providers operating in both markets should anticipate incident-reporting expectations under this channel and review internal incident-response procedures to be able to support government-to-government notifications.