U.S. Treasury Secretary Scott Bessent on Wednesday reiterated warnings to Chinese AI firms, saying sanctions remain a possibility after a White House official accused China-based Moonshot of improperly distilling Anthropic’s Fable model.
Model distillation is a common AI training technique in which a smaller model learns from the outputs of a larger one. While distillation can raise intellectual property concerns, it is also widely used as a legitimate optimization method.
Bessent posted on X: “Open source is not open season on American IP. When [Chinese] firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations will be on the table.” Earlier in the week he said the U.S. government would review Chinese open-source models for signs of IP theft and impose sanctions if such evidence is found.
Accusations involving GB300 hardware access
Bessent’s comments followed accusations from Michael Kratsios, the White House’s chief of science and technology policy, who said Moonshot conducted large-scale distillation against U.S. models. Kratsios alleged Moonshot had acquired Nvidia “GB300-equipped servers” and had accessed GB300s in Thailand, raising questions about whether the company violated U.S. export control rules.
The GB300 servers are part of Nvidia’s Blackwell generation, which are banned from sale to Chinese companies.
Doubts about a purely distillation-based origin for Kimi K3
Some experts dispute the claim that Moonshot’s Kimi K3 could have been developed primarily by distillation from Anthropic’s Fable. Anthropic’s Fable has been publicly available only since July 1. Moonshot released K3 last week as an open-weight model, and the model’s advanced capabilities have prompted questions about the business models of leading U.S. AI labs and whether they can continue to justify the large capital investments that underpin frontier AI development.
Broader debate in Washington over Chinese open models
The episode has intensified a broader debate in Washington about the influx of Chinese open models. Some, including Dean Ball, a former White House AI advisor and current OpenAI Head of Strategic Futures, have argued the U.S. should restrict or effectively ban the use of Chinese open-weight models to preserve America’s technological advantage and mitigate national security risks.
TechCrunch contacted Moonshot and the U.S. Treasury for comment, according to the reporting, and had not received responses at the time of publication.
