US Treasury threatens sanctions over alleged Fable distillation
Treasury chief Scott Bessent floats sanctions after the White House accused Moonshot AI of building Kimi K3 by distilling Anthropic's Fable model.
Washington's confrontation with China's open-weight model makers escalated on July 22, when Treasury Secretary Scott Bessent warned that sanctions are on the table after the White House accused Beijing-based Moonshot AI of building its Kimi K3 model by distilling Anthropic's Fable.
The accusations
Michael Kratsios, the White House's top science and technology official, alleged that Moonshot conducted large-scale distillation against US models — systematically querying them to train a rival system. He further claimed the company obtained servers equipped with Nvidia GB300 chips, which are barred from sale into China under US export controls, and accessed additional units in Thailand.
Bessent put the threat in unusually blunt terms, saying "open source is not open season on American IP," and indicated that sanctions and Entity List designations could follow if Chinese firms carry out what he called covert, industrial-scale distillation attacks that cross into IP theft.
A contested timeline
The technical case has drawn immediate skepticism. Fable only became publicly available on July 1, and Moonshot shipped K3 roughly a week later — a window several researchers say is too short for distillation to have played a primary role in the model's development. Neither government has published evidence so far, and the accusation lands amid a broader Washington debate, with some officials pushing to restrict or ban Chinese open-weight models outright.
Why it matters
This is the first time the US government has named a specific Chinese lab and a specific American model in a sanctions threat, moving the distillation dispute from rhetoric toward enforcement. If Treasury follows through, it would establish model weights as sanctionable intellectual property — a precedent with sweeping consequences for the open-source AI ecosystem, for Chinese labs' access to US model APIs, and for every developer who builds on Chinese open-weight models. The weak public evidence also raises the stakes: a sanctions regime built on a contested technical claim would deepen mistrust on both sides of an already fracturing AI landscape.