White House says China runs industrial-scale AI copying

The White House says foreign groups, mainly in China, run large distillation campaigns using thousands of proxy accounts and jailbreak techniques to copy U.S. AI models.

On Thursday the White House released a memorandum titled “Adversarial Distillation of American AI Models” outlining what it describes as coordinated efforts by foreign entities, primarily in China, to extract behavior and proprietary data from U.S.-based AI systems. Michael Kratsios, Assistant to the President for Science and Technology, wrote on X that the U.S. has evidence of “industrial-scale distillation campaigns.”

The memo defines a distillation attack as training a smaller model on the outputs of a larger one by repeatedly querying the target system. According to the document, attackers use large networks of proxy or fake accounts to avoid detection and apply jailbreaking techniques to bypass safety filters and access guarded capabilities. The administration says these campaigns can reproduce performance on selected benchmarks at a fraction of the original cost and can remove safety controls.

The memo cites a February incident in which Anthropic alleged that Chinese labs used roughly 24,000 fraudulent accounts to extract millions of responses from its Claude model and use them to train competing systems. The White House used that and similar incidents to illustrate the scale and method of the activity it is flagging.

The memorandum notes that models produced by unauthorized distillation may not match full original performance, but can match common tests and may lack built-in safety and content controls, raising the risk of harmful or biased outputs.

Federal agencies will work with U.S. AI firms to strengthen protections around frontier models. The administration plans to coordinate with industry on technical defenses, improve detection of coordinated account networks, expand information-sharing, and explore ways to hold foreign actors accountable. The memo does not list specific sanctions or enforcement measures.

The Office of Science and Technology Policy published the memo amid concerns over international competition in advanced AI and the security of American intellectual property. The document acknowledges lawful distillation can yield smaller, efficient open-source models but draws a line at systematic, unauthorized copying. The memo states, “There is nothing innovative about systematically extracting and copying the innovations of American industry.”

The material on GNcrypto is intended solely for informational use and must not be regarded as financial advice. We make every effort to keep the content accurate and current, but we cannot warrant its precision, completeness, or reliability. GNcrypto does not take responsibility for any mistakes, omissions, or financial losses resulting from reliance on this information. Any actions you take based on this content are done at your own risk. Always conduct independent research and seek guidance from a qualified specialist. For further details, please review our Terms, Privacy Policy and Disclaimers.

Articles by this author