Michael Kratsios, the White House director of science and technology policy, said in a memo Thursday that foreign companies, primarily based in China, are conducting an “industrial-scale” campaign to steal cutting-edge AI models from U.S. companies.
The efforts include the use of tens of thousands of proxy accounts to avoid detection and the use of complex tools to expose sensitive information, Krastios wrote.
“These coordinated campaigns systematically extract capabilities from American AI models and leverage American expertise and innovation,” he said.
The United States and China are locked in a race to build the most sophisticated and cutting-edge AI technology. AI has become a flashpoint in trade tensions between the two global superpowers, and tech giants like Nvidia are caught in the middle.
At the heart of the accusations is a process called “distillation,” a technique used to transfer knowledge from large-scale AI models to smaller-scale AI models that are cheaper to run. Distillation is widely used to train AI models, but companies like Anthropic and OpenAI have previously claimed that the process is being used to unfairly mimic the functionality of their models.
DeepSeek, the Chinese AI startup that caused a stir on Wall Street last year, is at the center of such accusations. OpenAI warned lawmakers in February that DeepSeek was attempting to replicate the performance of the ChatGPT maker’s model and other US Frontier Institute models through distillation.
Anthropic also announced in February that it had “identified an industrial-scale campaign” by Deep Seek and two other AI labs to “illegitimately extract Claude’s abilities to improve their own models.”
DeepSeek did not immediately respond to CNN’s request for comment.
Extracted AI models also pose potential security concerns because they lack the safeguards of the models used to train them, Anthropic said in a February blog post.
“These distillation campaigns also allow attackers to intentionally remove security protocols from the resulting models and reinstate mechanisms that ensure that AI models are ideologically neutral and truth-seeking,” Kratsios wrote in the note.
CNN has reached out to the Chinese embassy in Washington for comment.
The Trump administration plans to take several steps to counter this campaign, including sharing information with U.S. AI companies about attempts to conduct “unauthorized industrial-scale distillation” and strengthening collaboration among private companies.
It also plans to consider measures to “hold foreign actors accountable” for such campaigns and work with the private sector to develop best practices for detecting and defending against “industrial-scale distillation activities.”
Consolidating the United States as a leader in the AI race is a cornerstone of President Trump’s second term. He is pushing for federal AI regulation rather than state-level rules to encourage innovation. But critics warn that this approach could make it easier for AI companies to evade accountability. He has also exercised control over sales of AI chips to China.
