Key facts
- The U.S. and China are planning official bilateral discussions on AI safety risks.
- The talks are tentatively scheduled for mid-September, ahead of a U.S.-China leaders' summit on September 24.
- U.S. Treasury Secretary Scott Bessent is expected to lead the U.S. delegation.
- The U.S. intends to discuss cooperation on monitoring AI-directed cyberattacks and has proposed self-policing by AI labs.
- Concerns include the potential for advanced AI models to conduct cyberattacks and the alleged distillation of U.S. AI models by Chinese firms.
The United States and China are preparing to hold official bilateral discussions focused on AI safety risks, with the talks tentatively scheduled for mid-September. This dialogue comes as rapidly advancing AI capabilities reach a global tipping point, raising concerns about potential misuse and security threats.
The discussions are expected to be led by U.S. Treasury Secretary Scott Bessent on the American side. The U.S. aims to explore cooperation on monitoring AI-directed cyberattacks and has proposed that AI labs in both countries "police themselves" and share information to prevent such attacks. Washington is also concerned about the potential for future Chinese AI models to conduct sophisticated cyberattacks and has raised allegations of Chinese firms distilling proprietary U.S. AI models to lower development costs.
Chinese officials reportedly view these AI talks as a significant outcome for the upcoming summit between President Xi Jinping and U.S. President Donald Trump, scheduled for September 24 in Washington. China's cyberspace regulator has publicly warned about "extreme AI loss of control risks," and there is emphasis on ensuring any limits on frontier AI apply equally to models from both nations. The U.S. has also urged G20 members to adopt a hands-off approach to AI regulation, a stance China has aligned with by signing the Carolina Principles.
Recent incidents, such as nearly 700 rogue AI agents hacking Hugging Face and a swarm of AI agents hijacking a German website, underscore the urgency of addressing AI risks. Experts suggest that while outcomes may be limited, establishing a channel for sharing observations and jointly monitoring AI safety incidents would be a crucial starting point for managing cross-border crises.