The Trump administration has announced a significant initiative to combat the exploitation of American artificial intelligence models by foreign technology companies, with a particular focus on China. Michael Kratsios, the president’s chief science and technology adviser, issued a memo accusing foreign entities, primarily based in China, of systematic, industrial-scale efforts to “distill” or extract capabilities from leading U.S.-made AI systems. These actions, according to Kratsios, represent an exploitation of American expertise and innovation. The administration plans to collaborate with American AI companies to identify such illicit activities, bolster defenses, and implement punitive measures against offenders.

This crackdown emerges at a critical juncture as China increasingly challenges the United States' long-held dominance in artificial intelligence. The White House emphasizes the necessity of U.S. leadership in AI to establish global standards and secure economic and military advantages. However, a recent report from Stanford University’s Institute for Human-Centered AI indicates that the performance gap between top U.S. and Chinese AI models has "effectively closed," intensifying the strategic competition between the two nations.

In response to these allegations, Liu Pengyu, spokesperson for China’s embassy in Washington, voiced opposition to what he described as “the unjustified suppression of Chinese companies by the U.S.” Pengyu asserted China's commitment to advancing scientific and technological progress through cooperation and healthy competition, while also emphasizing the country's dedication to protecting intellectual property rights.

Further underscoring the severity of the issue, Kratsios’ memo coincided with the unanimous, bipartisan approval by the House Foreign Affairs Committee of a bill designed to establish a mechanism for identifying foreign actors who extract "key technical features" from closed-source, U.S.-owned AI models. The bill proposes punitive measures, including sanctions, against these offenders. Rep. Bill Huizenga, R-Mich., the bill’s sponsor, characterized "model extraction attacks" as "the latest frontier of Chinese economic coercion and theft of U.S. intellectual property," stressing the importance of preventing China from stealing transformative cyber capabilities derived from American AI models.

Specific instances of alleged exploitation have fueled these concerns. Last year, the Chinese startup DeepSeek drew attention in U.S. markets after releasing a large language model that rivaled U.S. AI giants in capability but at a significantly lower cost. David Sacks, then an AI and crypto adviser to President Donald Trump, suggested that DeepSeek had copied U.S. models, stating there was "substantial evidence" that DeepSeek "distilled the knowledge out of OpenAI’s models."

OpenAI, the creator of ChatGPT, echoed these concerns in a February letter to U.S. lawmakers, alleging that China should not be permitted to advance "autocratic AI" through the "appropriating and repackaging American innovation." Similarly, Anthropic, the developer of the Claude chatbot, accused DeepSeek and two other China-based AI laboratories in February of engaging in campaigns to "illicitly extract Claude’s capabilities to improve their own models." Anthropic detailed that this was achieved through the "distillation technique," which involves training a less capable model on the outputs of a stronger one. While acknowledging that distillation can be a legitimate method for AI training, Anthropic highlighted it as problematic when competitors use it to "acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently."

However, the flow of technological inspiration is not exclusively one-sided. San Francisco-based startup Anysphere, known for its coding tool Cursor, recently admitted that its latest product was based on an open-source model developed by the Chinese company Moonshot AI, creator of the chatbot Kimi. This example underscores the complex and often interdependent nature of global AI development.

Identifying and combating unauthorized distillation presents significant challenges. Kyle Chan, a fellow at The Brookings Institution and an expert on China’s technology development, likened the task to "looking for needles in an enormous haystack," given the vast volume of legitimate requests for data. Chan suggested that enhanced information sharing and coordination among U.S. AI laboratories, facilitated by the federal government, could play a crucial role in strengthening anti-distillation efforts across the industry. The ultimate trajectory of the proposed House bill remains uncertain, with Chan noting that President Trump might be hesitant to disrupt relations with Chinese President Xi Jinping ahead of a planned state visit to Beijing in mid-May.