Skip to main content

US Treasury Threatens Sanctions Against Moonshot After White House Exposes Fable Model Distillation

US Treasury Secretary Scott Bessent issues a warning to Chinese AI firms after the White House accuses Moonshot of secretly distilling Anthropic's Fable model.

AI-written
Inewgen
23 Jul 2026Source: TechCrunch3 min read (0 views)Last updated 27 Jul 2026
Share
US Treasury Threatens Sanctions Against Moonshot After White House Exposes Fable Model Distillation

Stock photo for illustration only, not from the actual event

Font size
  • US Treasury Secretary Scott Bessent threatens sanctions against Chinese AI companies
  • White House accuses Moonshot of secretly distilling Anthropic's Fable model
  • Michael Kratsios states Moonshot accessed Nvidia GB300 servers in Thailand

The U.S. Department of Treasury made a significant move this past Wednesday as Treasury Secretary Scott Bessent issued a stern warning to artificial intelligence companies from China, stating that sanctions remain an option the U.S. government is ready to deploy. This followed allegations from high-ranking White House officials that Moonshot had engaged in the unauthorized model distillation of Anthropic's Fable model.

Model distillation is a common AI training technique where a smaller model learns from the outputs of a larger model. Although widely used as a method to improve efficiency, this process can also potentially infringe upon intellectual property rights. Scott Bessent posted on the X platform that open source does not mean open season on American intellectual property, and if any Chinese company carries out distillation attacks at an industrial scale that cross the line into intellectual property theft, sanctions and Entity List blacklisting will definitely be considered.

"Open source is not open season on American IP."

Scott Bessent

Just days prior, Scott Bessent had stated that the U.S. government would scrutinize open-source models from China for signs of intellectual property theft and is ready to enforce punitive measures if wrongdoing is found. This latest warning came mere hours after Michael Kratsios, the White House head of science and technology policy, accused the Chinese company Moonshot of conducting large-scale distillation on U.S. models. He also alleged that Moonshot had purchased servers equipped with Nvidia GB300 chips and accessed these GB300 chips in Thailand, which could potentially violate U.S. export control regulations since Nvidia's Blackwell family servers are banned from being sold to Chinese companies.

Never miss the latest news?

Subscribe to get news summaries by email - not often enough to be annoying.

โฆษณา

artificial intelligence server hardware chips

Model distillation is a technique that enables AI development companies to quickly build small, highly cost-effective models that maintain performance close to larger frontier models. This dispute reflects U.S. concerns over the vulnerabilities of open-weight and open-source models, which allow competitors to access model weights and build upon them without massive historical investments, thereby undermining America's competitive edge on the global AI stage.

However, some experts have pushed back against the notion that Moonshot's Kimi K3 was primarily developed through distillation from the Fable model, given that Fable was only made publicly available on July 1st, whereas Moonshot launched K3 as an open-weight model the previous week. The advanced capabilities of K3 have raised questions regarding the business models of leading AI labs in the U.S. and whether they can continue to shoulder the massive capital investments necessary to compete at the frontier of AI.

The incident has also sparked widespread debate in Washington, D.C., regarding the influx of open-source models from China. Prominent figures such as Dean Ball, a former White House AI advisor and current head of future strategy at OpenAI, suggested that the U.S. should seriously restrict or ban the use of Chinese open-source models to preserve the nation's technological advantage and mitigate potential national security risks.

Source: TechCrunch

Comments

Leave a Comment
0/2000

Found something wrong in this article? Report an issue with this article