Two sources told Reuters that the United States and China are preparing to hold a dialogue in mid-September to discuss artificial intelligence (AI) security risks. The importance of this dialogue is growing as frontier AI capabilities develop rapidly and reach a critical global turning point.
Sources familiar with the matter revealed that this meeting will be the first official bilateral talk between the U.S. and China specifically focused on artificial intelligence since President Trump began his second term. The sources, who requested anonymity, stated that the U.S. side will be led by Treasury Secretary Bessent.
Reuters was the first to report the tentative timing of mid-September. A White House official stated, "There are currently no plans to hold an AI-related meeting in mid-September."
One of the anonymous sources said that the U.S. side hopes to discuss cooperation in monitoring AI-driven cyberattacks and has proposed a suggestion for "self-regulation" by Chinese and American AI laboratories, involving the sharing of relevant information to prevent AI-related cyberattack incidents.
The source stated that Washington is also concerned about the potential emergence in China of AI models with "Mythos-level" capabilities capable of launching cyberattacks, and plans to raise issues regarding China's alleged distillation of proprietary U.S. AI models.
The U.S. AI company Anthropic developed the "Mythos" model.
Distillation refers to using the output results of large, high-cost models to train smaller-scale models, thereby reducing the cost of training powerful new models.
Reuters was the first to disclose details related to the proposed agenda for these talks.
Another anonymous source indicated that the Chinese side might be led by Vice Premier He Lifeng, who is Bessent's primary counterpart on U.S.-China economic affairs, but other senior officials such as ****, a member of the Standing Committee of the Political Bureau of the CPC Central Committee, might also be considered to lead the delegation.
Reuters could not confirm the location of the talks or the identities of other attendees, but the second source added that White House Technology Advisor Michael Kratsios and Chinese Minister of Science and Technology Yin Hejun might attend the meeting.
Sources indicated that the final agenda and list of attendees are still being adjusted.
The White House declined to comment on the artificial intelligence talks. The U.S. State Department did not respond to requests for comment. China’s Ministry of Foreign Affairs, Ministry of Science and Technology, Ministry of Industry and Information Technology, National Development and Reform Commission, and cyberspace regulators also did not immediately respond to requests for comment.
Kratsios’s office did not immediately respond to requests for comment.
Editor/Rocky