The White House’s top tech policy official, Michael Kratsios, is accusing Moonshot AI — the Chinese company behind the Kimi chatbot — of training its latest model, Kimi K3, by distilling Anthropic’s Fable. (Distillation is the practice of using a stronger model’s output to train a smaller one.) Treasury Secretary Scott Bessent subsequently said sanctions and placement on the Entity List are under consideration.
Kratsios claims Moonshot AI built an internal platform that can switch between multiple access channels to evade detection. He also alleges the company purchased servers equipped with Nvidia’s GB300 chips and is using them in Thailand. Kratsios did not provide public evidence for these claims.
Anthropic itself previously accused Moonshot AI in February of using hundreds of fake accounts to interact with Claude over 3.4 million times, specifically targeting agentic reasoning, tool use, coding, and data analysis capabilities. Moonshot AI did not respond to media requests at the time.
Moonshot AI has not yet responded to the latest accusations. The Chinese embassy in the US called the claims “completely unfounded.” Inside the US government, there is not yet a unified approach — the White House is pushing for tighter restrictions, while the Commerce Department argues that a blanket ban would be difficult to enforce.
https://twitter.com/mkratsios47/status/2079933645888880708?s=46