U.S. and Chinese Scholars Agree on Need to Curb Frontier AI Cybersecurity Risks and Systemic Loss of Control

The United States continues to restrict exports of advanced chips and models to China on national security grounds, while China is accelerating its pursuit through open-weight models including Moonshot AI's Kimi, Alibaba's Qwen and Z.ai's GLM. Agentic AI can autonomously find vulnerabilities, write code and conduct social engineering, allowing risks to spread across borders and making them difficult for either country to manage alone.

The Beijing Academy of Artificial Intelligence held its eighth BAAI Conference at the Zhongguancun International Innovation Center on June 12–13, 2026, bringing together more than 200 experts. MIT scholar Stephen Casper urged the industry to avert an AI “Chernobyl moment.” Shanghai Jiao Tong University professor Lin Yun called for joint safety standards and warned that GLM 5.2 already possessed frontier-level agentic and coding capabilities.