AI Tech / news brief

U.S. Treasury Warns of Sanctions After Chinese AI Firm Moonshot Accused of Distilling Anthropic’s Model

U.S. Treasury Secretary Scott Bessent said sanctions remain on the table after a White House official accused Chinese AI company Moonshot of improperly distilling Anthropic’s Fable model, escalating tensions over intellectual property and export controls.

U.S. Treasury Secretary Scott Bessent doubled down on warnings to Chinese artificial intelligence companies on Wednesday, stating that sanctions and Entity List designations could be imposed for intellectual property theft. The remarks came after a White House official accused China-based Moonshot of conducting large-scale distillation of Anthropic’s Fable model.

Model distillation is a common AI technique where a smaller model learns from the outputs of a larger one. While it can be used legitimately for optimization, officials argue that covert, industrial-scale distillation crosses the line into IP theft.

“Open source is not open season on American IP,” Bessent posted on X. He said that when Chinese firms conduct such attacks, sanctions will be considered.

Earlier this week, Bessent stated the U.S. government would examine Chinese open-source models for signs of IP theft. The latest escalation follows an accusation by White House science and technology policy chief Michael Kratsios, who alleged Moonshot had acquired Nvidia’s GB300-equipped servers and accessed GB300s in Thailand, likely to train AI models. The GB300 servers, part of Nvidia’s Blackwell generation, are banned for sale to Chinese companies, raising potential export-control violations.

The episode has intensified a broader debate in Washington over the influx of Chinese open-source AI models and their compliance with U.S. export rules.