The Cyberspace Administration of China has launched an inquiry into AI firms DeepSeek and Moonshot AI following allegations that they diverted sensitive military and police information to Anthropic’s Claude. The probe focuses on whether the companies routed data from Chinese government users and state-owned corporations to the U.S. without the knowledge of their customers.

An Anthropic report detailed how some Chinese users utilized transfer stations to covertly access advanced AI models, accidentally sending information to American servers. This practice undermines U.S. efforts to restrict high-end AI technology from being accessed within China.