US and China Reportedly Plan Talks on AI Safety

The United States and China may soon sit down to discuss one of the defining fears surrounding artificial intelligence: how to contain its security risks. Reuters reports that the two powers could hold talks in mid-September, although the White House has not confirmed the plan and China has made no corresponding announcement.

US Treasury Secretary Scott Bessent is expected to lead the American side. He could face He Lifeng, who oversees China’s economic and financial policy. The timing would fit with a planned September 24 meeting in Washington between US President Donald Trump and Chinese President Xi Jinping, though Reuters could not determine where the AI discussions would take place.

One source told Reuters that the countries could discuss cooperation in monitoring AI-driven cyberattacks. The United States has also reportedly proposed asking American and Chinese AI laboratories to “self-regulate” and share information to help prevent cyberattacks linked to artificial intelligence.

Washington is also expected to raise a separate complaint: that Chinese companies are excessively imitating US AI services by using American models and their outputs as training material after those systems were developed with substantial money and effort.

The reported talks come amid broader international debate over AI safety and the growing capabilities of modern tools. Companies such as OpenAI and Anthropic have stressed the importance of information security while unveiling increasingly advanced models, developments that have also fueled concern about more autonomous AI systems.