China May Be Following US Lead With Quiet Crackdown on AI Exports
China is reportedly considering new limits on access to its advanced AI models after the U.S. briefly restricted Anthropic models in June. Reuters says Chinese officials met with Alibaba, ByteDance, and Z.ai to discuss restricting overseas use of both closed-source and open-weight models, including future releases, and possibly tightening rules on AI-related foreign investment and unauthorized tech disclosure under national security law. Any policy is still undecided and may apply only to future models. A proposed framework would treat basic open-source tools lightly, require security review for more advanced systems, and bar or domestically confine the most sensitive frontier models. Such a move would reverse the openness that helped Chinese models gain global users. It could raise costs for developers and remove a major alternative to U.S. systems. The talks also reflect China’s growing concern that foreign AI tools, especially cybersecurity models, could be used against Chinese infrastructure.
