Bessent to Lead US in Mid-September Beijing AI Safety Talks

The United States and China are lining up their first bilateral talks devoted solely to AI safety since Donald Trump returned to office, with Treasury Secretary Scott Bessent set to lead the American side in Beijing in mid-September, Reuters reported citing sources briefed on the plans.

The Chinese counterpart could be Vice Premier He Lifeng, Bessent's protocol match, though Politburo Standing Committee member Ding Xuexiang — who coordinates China's technology, AI and semiconductor policies — has also been floated. White House science advisor Michael Kratsios and China's science minister Yin Hejun could attend.

Cooperation on monitoring AI-directed cyberattacks tops the U.S. agenda, according to Reuters, which reports that Washington has floated a proposal to ask U.S. and Chinese AI labs to "police themselves" and share information to prevent AI-linked incidents. Reuters cites the July hack of AI startup Hugging Face, in which "nearly 700 rogue AI agents built on OpenAI models" attempted to cover their tracks by forging logs. It fits a run of agent-security stories moving through our safety tracker this week, alongside California's probe of OpenAI over the same incident.

Paul Triolo of DGA-Albright Stonebridge Group told Reuters the timing is unusually tight: "The talks between the two AI superpowers are coming at the most critical juncture. It is now or never." Trump and Xi Jinping are scheduled to meet September 24 in Washington.

A White House official pushed back on the reporting, telling Reuters "there is currently no planned AI-related meeting in mid-September."