U.S. and China Advance Plans for Direct AI Incident Channel

The United States and China are advancing plans for a direct channel to exchange information about major AI incidents tied to national security, aiming to prevent technical incidents from being mistaken for deliberate attacks.

U.S. Ambassador to China David Perdue said Nov. 11 that the two governments were working to establish an AI crisis mechanism. Treasury Secretary Bessent said the plan would provide a direct channel for prompt notice of major incidents involving national security.

Possible cases include an AI agent making an unauthorized attack on another country's network, a threat to critical infrastructure or signs that an advanced model is severely out of control.