According to a recent Reuters report, the U.S. State Department has instructed its missions worldwide to amplify claims that Chinese companies "steal" U.S. AI intellectual property through so-called "distillation," laying the groundwork for potential follow-up actions and coordinated international messaging by Washington.

On April 23, Michael Kratsios, the chief science and technology adviser to President Donald Trump, issued a memorandum accusing Chinese entities of using "distillation" to "extract capabilities from American AI models, exploiting American expertise and innovation." The memo claimed that "distillation" allows foreign entities to release seemingly comparable products at very low cost. On the same day, the U.S. House Committee on Foreign Affairs introduced a bill calling for the identification and sanctioning of foreign entities that acquire capabilities from U.S. AI models via "distillation."

These moves suggest a coordinated push within U.S. political circles to elevate the narrative that China is leveraging "distillation" to appropriate American AI technology. The objectives appear twofold: to stigmatize China's AI development under the banner of intellectual property violations, thereby weakening its global influence; and to create a policy rationale for expanded sanctions on Chinese AI companies. Similar tactics have been used in recent years to restrict China's broader technology sector.

Yet the claim that China's AI progress hinges on "distillation" or the misappropriation of U.S. technology is difficult to sustain. Many Chinese AI companies have adopted relatively open development paths, publishing research and making technical approaches accessible. Their advances in core large-model architectures and training methods have been widely recognized in both industry and academia. Against this backdrop, such sweeping allegations appear more  political than technical.

Take DeepSeek as an example. In January 2025, the company released its DeepSeek-R1 model. Around the same time, OpenAI was reported to have suggested to media outlets that DeepSeek may have replicated U.S. capabilities through "distillation."

That claim was soon challenged by publicly available research. In September 2024, DeepSeek published a paper in Nature detailing the core algorithms behind R1, making it one of the first large-scale machine learning models to be described in a peer-reviewed setting. One of the reviewers, machine learning engineer Lewis Tunstall, noted that DeepSeek's reasoning methods were sufficiently advanced and did not rely on distilling OpenAI's systems.

On April 24, DeepSeek unveiled its next-generation model, DeepSeek-V4, capable of handling context windows of up to one million Chinese characters. Earlier research on long-context processing by the company stood out among more than 8,300 submissions and received the Best Paper Award at the 2025 annual meeting of the Association for Computational Linguistics (ACL), one of the field's leading international conferences.

Such academic outputs underscore a key trend: China's AI development is increasingly driven by indigenous innovation, a trajectory closely observed by the global research community.

In contrast, many U.S. AI companies favor closed-source approaches and disclose minimal technical detail. The extent to which similar techniques — such as distillation — are used in their own development processes, or whether cross-model influences run in multiple directions, are questions that receive far less public scrutiny.

In February, U.S. AI firm Anthropic accused several Chinese companies of distilling its models. Ironically, around the same time, researchers reported that in certain cases Anthropic's model, when queried in Chinese about its identity, responded that it was developed by DeepSeek.

At the 2025 ACL annual meeting, more than half of first authors on accepted papers were affiliated with Chinese institutions. As the global AI landscape evolves, China has emerged as a major center of R&D. China's AI progress relies on the expertise and innovation of its scientists — something the U.S. cannot contain or suppress. No matter how much they hype it up, they are only fooling themselves.