U.S. proposes AI emergency alert system with China: What to know

In the latest update, Treasury Secretary Scott Bessent said during a television appearance that he is proposing a communication workflow between the US and China if something goes wrong with AI, on the assumption that Beijing will agree.

The primary motive behind Bessent’s idea reflects what he calls China’s sudden epiphany about the risks of its open-source AI posing a more significant threat than Beijing had realized.

Bessent and Chinese Vice Premier He Lifeng discussed AI safety ahead of Xi Jinping's state visit last month featuring a proposed AI incident reporting workflow. Speaking to Axios, Bessent covered risks such as AI agents and malicious actors using technology to create cybersecurity threats.

In this connection, Bessent said: “I think that they didn't realize how powerful their open source models are.”

China is already concerned about a different kind of AI threat that the technology could weaken the Communist Party’s tight control over information.

Earlier, China’s spy chief warned that AI could be used to threaten political and ideological stability. The US is also navigating the AI safety dilemma with rising concerns about the technology’s risks. Bessent advocated for slowing down increasingly powerful models, stating that the US needs to focus more on protecting crucial systems from AI threats rather than just focusing on making models more powerful.