The White House has accused Chinese AI startup Moonshot AI of stealing Anthropic’s recently launched Fable model while developing Kimi K3, one of the world’s most advanced open-weight artificial intelligence models. The allegations mark the first time the US government has publicly backed claims that a Chinese AI company copied a leading American model.
Michael Kratsios, director of the White House Office of Science and Technology Policy, said the US government has information indicating Moonshot carried out large-scale “distillation” of Anthropic’s Fable model. Distillation is an AI training technique where one model learns by analysing another model’s responses. While the method is widely used across the industry, Kratsios alleged Moonshot built internal systems to carry out the process covertly and at a scale that amounted to theft of proprietary technology.
Kratsios also alleged that Moonshot acquired servers powered by Nvidia’s latest GB300 Blackwell chips and deployed them in Thailand despite US export restrictions that prohibit Chinese firms from accessing the advanced hardware.
The accusations came just days after Moonshot unveiled Kimi K3, a 2.8 trillion parameter open-weight model that rivals leading systems from Anthropic and OpenAI on several coding and reasoning benchmarks. The company plans to publicly release the model’s weights on July 27, allowing developers worldwide to download and modify the system.
Anthropic has previously accused Moonshot, DeepSeek, MiniMax and Alibaba of using its AI models without permission. However, those allegations had largely remained a dispute between AI companies. The White House’s intervention significantly raises the stakes, with Treasury Secretary Scott Bessent also saying the US could impose sanctions on Chinese companies found to have stolen American AI technology.
The White House has not released evidence supporting its allegations, and Moonshot has not responded publicly. Some AI researchers have also questioned whether the timeline supports the claims, noting that Anthropic released Fable only about a week before Kimi K3.
Regardless of how the allegations unfold, the episode signals that disputes over frontier AI models are increasingly becoming matters of government policy, placing AI models alongside advanced semiconductors as technologies the United States is prepared to defend.
