Anthropic claims that China’s Alibaba used 25,000 fake accounts and 28.8 million exchanges to illicitly ‘distill’ its Claude model — violations occurred from April to June 2026

Anthropic, the American AI lab that created one of the most advanced large language models available today, has said in a letter to the U.S. Senate that Chinese AI tech giant Alibaba has illicitly used Claude to train its own models. According to Reuters, the company sent the letter to Sen. Tim Scott (R-S.C.), the chair of the U.S. Senate Banking Committee, and Sen. Elizabeth Warren (D-Mass.), the ranking member of the said committee, before a scheduled hearing set to tackle AI issues.

This isn’t the first time that Anthropic has accused Chinese AI labs of “stealing” the capabilities of its AI model to train their own. Earlier this year, the company claimed that DeepSeek, Moonshot, and MiniMax used 24,000 fraudulent accounts and made 16 million exchanges collectively to train their own AI LLMs on Claude’s output. This method of training AI using the output of a more advanced model is called distillation, and while there are legitimate uses of this technique, such as when a frontier AI model is distilled to create a lighter, cheaper version of itself, it argues that competing labs can also use the same technique to build their own models at “a fraction of the time, and at a fraction of the cost, that it would take to develop them independently.”

The American AI lab says that it traced the distillation campaign back to operators that had connections with Alibaba, one of the largest Chinese tech companies often compared to Amazon, and Alibaba Qwen, its AI lab. Anthropic warns that distillation may help China create a frontier AI model that could match Mythos Preview’s capabilities — something many American lawmakers are afraid of.

While U.S. tech companies still enjoy an advantage when it comes to the latest AI models, Chinese tech companies are quickly catching up. In fact, Elon Musk estimated that a Chinese AI lab would have achieved a Fable 5-class AI model by the first quarter of next year, but the CEO and founder of Chinese AI lab Z.ai confidently replied, “won’t take that long.” Aside from that, many enterprise users are slowly switching to more affordable open-source Chinese LLMs for their agents as token costs spiral out of control, reserving the most powerful (and expensive) American models only for the most complicated tasks.

Both the U.S. and China are pushing hard to achieve AI supremacy, with the two rivals taking steps to reduce the advantage of the other. For example, Washington has been using export controls to limit Chinese access to advanced hardware needed to build the most advanced chips and for training AI, while Beijing countered this with its own controls on rare earth materials, some of which serve as key ingredients in chip making.