White House: China-Linked AI Theft Tactics Exposed

WASHINGTON — The White House said on Tuesday it plans to deepen coordination with U.S. artificial intelligence companies in response to what it described as “industrial-scale campaigns” by foreign entities, primarily based in China, aimed at stealing AI advances.

Michael Kratsios, Director of Science and Technology Policy, outlined the administration’s position in an internal memo, saying new intelligence points to foreign actors exploiting American firms through a method referred to as “distilling.” In the memo, this was described as copying AI technology developed by U.S. companies.

According to the memo, the alleged distillation efforts involve foreign firms running large numbers of individual accounts that interact with AI chatbots or other tools while presenting as ordinary users. The memo said these accounts then attempt to “jailbreak” systems or otherwise expose proprietary information about AI models. The information obtained is described as being used to support the development and training of the foreign actors’ own AI models.

The White House said it intends to provide U.S. AI companies with additional information about the tactics described in the memo. It also plans to coordinate efforts to counter such attacks, develop best practices to reduce exposure, and explore accountability mechanisms for foreign actors.

The memo did not specify what actions, if any, would be taken directly against foreign entities. It also did not provide details on particular companies or organizations alleged to be involved in the activity described.

A representative from China’s U.S. embassy in Washington, D.C., rejected the White House’s claims. The representative said China’s AI progress comes from its own work and international cooperation, and criticized what they called “the unjustified suppression of Chinese companies by the US.”