U.S. and China Agree to AI Safety Protocol at Trump-Xi Summit in Beijing

The United States and China agreed Thursday to establish a protocol on best practices for artificial intelligence safety, a rare area of cooperation announced on the sidelines of President Donald Trump's two-day summit with Chinese President Xi Jinping in Beijing.

U.S. Treasury Secretary Scott Bessent disclosed the agreement in an interview with CNBC's Joe Kernen, describing it as an effort to prevent non-state actors from obtaining access to advanced AI models.

"The two AI superpowers are gonna start talking. We're gonna set up a protocol in terms of how do we go forward with best practices for AI to make sure non-state actors don't get a hold of these models," Bessent told CNBC from the sidelines of the meeting.

Bessent framed the U.S. willingness to engage China on the topic as a function of American technological advantage. "The reason we are able to have wholesome discussions with the Chinese on AI is because we are in the lead," he said. "I do not think we would be having the same discussions if they were this far ahead of us."

The summit, held at the Temple of Heaven in Beijing, marks the first time a sitting U.S. president has visited China since 2017, when Trump also made the trip during his first term. The two leaders met Thursday morning, with the session wrapping up at 12 p.m. local time, and the summit is scheduled to conclude Friday.

Bessent also told CNBC he anticipates a significant "step-function jump" in upcoming large language model releases from Google's Gemini and OpenAI, underscoring how quickly the competitive landscape is shifting.

The AI agreement comes against a backdrop of ongoing U.S. efforts to limit China's access to advanced semiconductor technology. Washington has restricted exports of high-end chips, primarily those manufactured by Nvidia, to Chinese buyers. Nvidia CEO Jensen Huang joined Trump's delegation to China as a late addition to the trip.

When asked about a report that Washington had cleared sales of Nvidia's H200 AI chips to several major Chinese technology firms, Bessent said there had been "a lot of back and forth" on the matter, without providing further detail.

The discussions also touched on Taiwan. Beijing's readout of the Trump-Xi meeting said Xi emphasized that Taiwan is the most important issue in the bilateral relationship, warning that mishandling the matter could push the two countries toward conflict. Bessent told CNBC that Trump would address Taiwan "in the coming days," adding that "anyone who's been saying otherwise does not understand the negotiating style of Donald Trump."

The AI dimension of the summit also carries an industry dimension. U.S.-based Anthropic has drawn attention in Washington and abroad with its Mythos AI model, which is reported to have powerful cyberattack capabilities. The company said it would initially release the model to select business partners.

Ahead of the Trump-Xi meeting, Bessent met in Seoul with Chinese Vice Premier He Lifeng. China's Commerce Ministry characterized those preliminary talks as an effort to resolve trade issues and "further expand pragmatic cooperation." Bessent also visited Tokyo earlier in the week, where he discussed critical minerals and investment agreements with South Korean President Lee Jae Myung and Japanese Prime Minister Sanae Takaichi.

How the proposed AI safety protocol will be structured, governed, or enforced remains unclear, and both governments have yet to release formal documentation of the agreement. The scope and timeline for implementation will likely be among the key details to emerge as the summit concludes Friday.

Disclaimer