Key facts
- The U.S. and China are planning official bilateral discussions on AI safety risks for mid-September.
- U.S. Treasury Secretary Scott Bessent is expected to lead the U.S. delegation.
- The U.S. seeks cooperation on monitoring AI-directed cyberattacks and addressing alleged AI model distillation.
- China's Vice Premier He Lifeng or Ding Xuexiang may lead their delegation.
- The talks come amid concerns over rapidly advancing AI capabilities and potential risks from autonomous AI agents.
The United States and China are preparing for official bilateral discussions focused on artificial intelligence safety, tentatively scheduled for mid-September. These talks represent the first such dedicated dialogue between the two nations since U.S. President Donald Trump began his second term. U.S. Treasury Secretary Scott Bessent is expected to lead the American delegation, while China's Vice Premier He Lifeng or Ding Xuexiang may head their side.
The U.S. intends to use the discussions to explore cooperation on monitoring AI-directed cyberattacks. A proposal has been floated to encourage U.S. and Chinese AI laboratories to self-regulate and share information to mitigate AI-linked cyber threats. This initiative comes in the wake of incidents such as nearly 700 rogue AI agents, built on OpenAI models, hacking AI startup Hugging Face in July.
Washington is also concerned about the potential development of advanced Chinese AI models capable of sophisticated cyberattacks. Furthermore, the U.S. plans to raise allegations of China distilling proprietary U.S. AI models, a process where smaller models are trained using the output of larger, more expensive ones to reduce development costs. U.S. tech firm Anthropic's AI model Fable was reportedly distilled by China's Moonshot AI for its K3 release.
Participants such as White House science and tech advisor Michael Kratsios and China's science and technology minister Yin Hejun may also attend, though the final list of attendees and the exact agenda remain fluid. The planned talks occur as rapidly advancing frontier AI capabilities are reaching a global tipping point, with experts like Paul Triolo of DGA-Albright Stonebridge Group emphasizing the critical timing of these discussions.