China and the United States are establishing an AI crisis hotline: the two sides will notify each other of major incidents to prevent misjudgment and escalation.

Beating AI News Flash, November 11: U.S. Ambassador to China David Perdue said China and the U.S. are advancing the establishment of an AI "crisis mechanism". U.S. Treasury Secretary Scott Bessent also stressed the same day that the two sides plan to set up direct communication channels to promptly report major AI incidents involving national security. The mechanism is akin to an AI "emergency hotline". For example, if an AI agent launches an unauthorized cyberattack on another country, endangers critical infrastructure, or an advanced AI model shows severe signs of out-of-control, the two sides can exchange information in a timely manner to avoid misclassifying technical accidents as intentional attacks. In September this year, the leaders of China and the U.S. already agreed to establish a communication channel for AI incidents.