China’s Hottest AI Company Accused Of Stealing Anthropic Data In ‘Covert’ Plot

The Trump administration is accusing a Chinese artificial intelligence (AI) company of conducting a “covert” plot to steal Anthropic’s data to boost its AI models.

Moonshot AI, a Chinese AI lab, recently released its latest model, Kimi K3, which now threatens America’s predominance in the global AI race. Moonshot said it experienced so much demand immediately after Kimi K3’s release that it had to temporarily pause new subscriptions.

The Trump administration believes Anthropic’s flagship model was used to train Kimi K3 through a process known as “distilling.”

Moonshot developed a “large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection, Michael Kratsios, the White House director of the Office of Science and Technology Policy (OSTP), wrote Wednesday.

“Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable,” the Trump administration official stated.

The Chinese Embassy in Washington, D.C., referred the Daily Caller News Foundation to Foreign Ministry Spokesperson Lin Jian’s statement during Wednesday’s press briefing.

“The development of AI in China comes from greater self-reliance and strength in science and technology, and is fueled by China’s vision of extensive consultation and joint contribution for shared benefit,” the spokesperson said. “All countries should take a people-centered approach and develop AI for the positive and for good to ensure the open and inclusive development of AI for good and for all so as to better contribute to social progress and the common welfare of the international community. China opposes politicizing and instrumentalizing trade and tech issues. Such actions will only stifle global AI advances and serve no one’s interests.”

Kimi K3’s release follows as Yahoo Finance found that many companies in Silicon Valley, such as coding startup Cursor, have used Kimi to develop their product. Calle, a Bitcoin software developer, praised Kimi K3 for fixing security issues that Anthropic and OpenAI’s models would allegedly not fix due to “cyber guardrails.”

Moonshot AI did not respond to a request for comment.

“This is more than a heist of invaluable American Intellectual Property. It is an assault on every economy that prizes the entrepreneur, rewards private capital, and relies on fair and honest competition,” Jacob Helberg, the undersecretary of State, wrote Wednesday.

“Illicit, adversarial distillation is IP theft and industrial espionage that supports adversary military and intelligence capabilities. It is a national challenge that creates serious national security risks for the United States and democratic allies. We’ll continue to crack down on it and work with the Administration and Congress to maintain American AI leadership,” Sarah Heck, the head of public policy for Anthropic, wrote.

After the advent of Kimi K3, the Trump administration has become more hawkish towards Chinese AI labs after holding back from sanctioning Chinese AI firms for fear of stifling innovation.

Treasury Secretary Scott Bessent on Wednesday warned that, while the Trump administration supports open-source AI, the administration may sanction Chinese AI labs, or add to the Commerce Department’s “Entity List” that would close American access to Chinese AI models without a license.

Many lawmakers have scrutinized Chinese open-source models.

House Homeland Security Committee Chairman Andrew Garbarino and House Select Committee on China Chairman Rep. John Moolenaar in April announced a joint committee investigation into Chinese open-source models using distillation to boost their models via American AI models.

Dean Ball, the head of OpenAI’s strategic futures, argued on July 17 that the American government should create enough “regulatory risk” around using Chinese open-source models that companies would not use them.

China’s rapid AI development cannot be solely pinned on distillation, some experts say.

“I think it is mistaken to attribute all of the success of Chinese labs to distillation, but it’s just as much of a mistake to pretend like distillation doesn’t give Chinese labs a big advantage,” Ben Thompson, a business and technology analyst, wrote on his Substack, Stratechery.

Thompson argued the United States should pass legislation that would make collecting data for training AI models fair use and prohibit terms of service that ban distillation. He said that stopping distillation is “nearly impossible.”

“All of this is to say that I think the reaction to Kimi and Chinese models generally is pretty over-blown, at least from an economic perspective,” he added, arguing that Kimi K3’s performance does not mean Chinese models are cheaper to use than Anthropic or OpenAI’s models.

“Distillation does not explain China’s entire open-model lead. Chinese labs have world-class researchers, substantial compute, strong pre-trained models, software-hardware codesign, and rapidly improving post-training capabilities,” Dean Meyer, a partner at the Sequoia venture capital firm, wrote. “But distillation compresses the costly final gap between a strong base and a near-frontier system. Even if distillation represents a smaller share of a Chinese model’s total capability, it represents a meaningful share of its advantage over American open models.”

Kevin Bankston, a senior advisor on AI for the Center for Democracy and Technology, argued that distillation does not necessarily violate copyright.

“According to your (C) office, LLM outputs aren’t copyrightable, and courts have long held reverse engineering to be a fair use; trade secret is a stretch since the company’s model is directly disclosing the info,” he said, referring to the White Office of Science and Technology Policy.

“The labs objections are also, as widely pointed out, incredibly hypocritical, and as less widely pointed out, rather inconsistent with their fair use arguments in cases about their scraping data for AI training (which is why they haven’t yet articulated a legal theory about how what China’s doing is illegal, either.),” he added.

Anthropic in February alleged that Moonshot employed hundreds of fraudulent accounts to boost its AI models agentic reasoning, or the part of an AI agent that handles decision-making , coding and data analysis, and its computer-use agent development. The American AI company said that distillation can help develop AI models at a fraction of the cost it would take an AI lab to develop a model. Anthropic urged the AI industry and policymakers to work together to respond to these alleged distillation attacks.