Full Breakdown
U.S. Accuses Chinese AI Firms of Industrial-Scale Model Distillation; Beijing Denies Misconduct
By Drooid · · How we work
Accusations of Industrial-Scale AI Distillation
On September 9, U.S. The advisory, issued on September 8 by the National Security Agency, the Federal Bureau of Investigation and the Cybersecurity and Infrastructure Security Agency, alleged that the firms copied variants of American models from Anthropic, OpenAI, Alphabet’s Google and SpaceX to train their own products more quickly, “likely with Chinese government awareness.”
Technical Background of Distillation
Distillation is a machine-learning technique in which a smaller or newer model is trained on the outputs of a larger, more expensive model. Industry leaders such as Apple, Google and OpenAI have employed the method to transfer capability and reduce training expenses. Chinese officials describe the process as a “normal technical and commercial practice” used worldwide, arguing that it does not inherently constitute theft.
Responses from the United States and China
U.S. officials presented the allegations as part of a broader effort to protect American AI intellectual property ahead of a planned summit between President Donald Trump and Chinese President Xi Jinping later this month, as well as a mid-September dialogue on AI safety. China’s Ministry of Commerce issued a question-and-answer notice on September 9, calling the U.S. A foreign-ministry spokesperson urged the United States to honor the consensus reached by the two leaders and refrain from “false accusations and smearing China.”
Potential Security Implications
Reuters reported on July 31 that Chinese military researchers have used outputs from leading U.S. AI models to train domestic systems, suggesting a pathway for the technology to enhance defense capabilities. The U.S. advisory therefore links the commercial practice of distillation to possible military applications, while Chinese officials contend that the same technique is employed by American firms without restriction.
Next Steps in U.S.–China AI Dialogue
Both governments have signalled a willingness to discuss AI safety risks during the upcoming bilateral talks, but the dispute over distillation remains unresolved. The U.S. advisory concluded with defensive measures for American companies, whereas Beijing’s Ministry of Commerce demanded that Washington stop “politicising and weaponising” the practice, leaving the technical and legal boundaries of legitimate distillation undefined.
