U.S. officials are raising the stakes in the fight over Chinese open-weight AI models, with Treasury Secretary Scott Bessent saying sanctions and Entity List designations could be used when overseas firms cross from legitimate distillation into alleged intellectual-property theft.

What changed

The latest warning followed a public allegation from White House science and technology policy chief Michael Kratsios, who said the administration had information that Moonshot AI distilled Anthropic's Fable model while developing Kimi K3. Kratsios also alleged that Moonshot built an internal platform for large-scale distillation and accessed Nvidia GB300 systems in Thailand.

Bessent then posted that "open source is not open season on American IP," saying covert industrial-scale distillation attacks could trigger sanctions or Entity List action. That language is stronger than a general policy concern, but it is still a threat rather than an announced enforcement action.

Why it matters

The conservative read is that Washington is now signaling possible penalties against AI model developers themselves, not only chip suppliers, cloud intermediaries, or export-control violators. That would be a notable escalation for companies distributing open-weight models across global developer infrastructure.

The claim remains contested territory. Distillation can be a normal optimization technique, and TechCrunch reported that some experts dispute whether Kimi K3 could have been developed primarily from Fable, which became publicly available only weeks earlier. Moonshot has not publicly conceded the allegation. For developers, the practical risk is that model provenance, training access, and jurisdiction may become deployment questions alongside benchmark performance and price.