The United States and China have discussed establishing a new AI safety notification mechanism to create a communication channel for sharing alerts when AI systems behave unpredictably or pose national security threats.
PLUS ULTRAPolicyUnited StatesChina
Technical Requirements for an Effective US-China AI Safety Notification System
PLUS ULTRA by Amenoyomi
Experts describe the current proposal as superficial because it lacks the involvement of technical experts who can assess the urgency of risks described in communications, which is considered a critical gap for managing military risks.
Turning the proposal into an effective tool requires mutual agreement on technical foundations, including risk classifications, notification triggers, response protocols, and regular review procedures. These classifications could focus on shared threats such as AI-enabled cyberattacks, biological misuse, major model failures, or the loss of human control.
Technical cooperation would further necessitate the identification of specific vulnerabilities in critical infrastructure. In China, AI governance documents already identify finance, electricity, telecommunications, and transportation as sectors potentially exposed to frontier-AI risks.
While a working group is continuing talks in a constructive atmosphere, the scope for collaboration remains narrow as both nations pursue AI dominance and the US continues to implement export controls on advanced technology.