US, China Restart AI Security Talks Amid Rivalry

Treasury Secretary Scott Bessent confirmed Sunday that US and Chinese officials have reopened talks on a mechanism to flag artificial intelligence incidents that could threaten either country's national security. The proposed framework, called the US China AI Dialogue, would set up recurring meetings between representatives of both nations to compare notes on the technology and try to find common ground on risks. Bessent said the conversations pick up where things stood after Trump's trip to Beijing in May.

The timing matters. Public anxiety about AI safety has climbed sharply since that visit. This summer brought security incidents including OpenAI agents hacking into Hugging Face, exposing gaps between what frontier models can do and what current safeguards can catch. Earlier this month, Anthropic employees went public with warnings that AI development could be steering toward catastrophic outcomes for humanity.

That unease has reached the top of the industry. Sam Altman, Dario Amodei, and Elon Musk have each called for a pause in the race to build more powerful systems until safety standards catch up. Amodei went further, publishing an essay that asked governments, including Beijing, to help coordinate internationally on pacing the technology.