U.S. Treasury Secretary Scott Bessent doubled down on his warnings to Chinese AI companies, stating that sanctions remain a possibility after the White House accused Moonshot of improperly distilling Anthropic's Fable model.

The common practice of model distillation, where a smaller model learns from a larger one, has been scrutinized for its potential to infringe on intellectual property rights. However, it is also used as an optimization method. Bessent warned that when Chinese firms conduct 'covert, industrial-scale distillation attacks,' sanctions and Entity List designations will be considered.

Earlier this week, Bessent stated that the U.S. government would examine open source models from China for signs of intellectual property theft and impose sanctions if found. The accusation comes hours after Michael Kratsios, White House science and technology policy chief, accused Moonshot of using Nvidia’s GB300-equipped servers to train its AI models in Thailand.

Some experts dispute the claim that Kimi K3 could have been developed primarily through distillation from Fable. K3 was released by Moonshot last week as an open-weight model and has advanced capabilities, raising questions about leading U.S. AI labs’ business models and justifying their enormous capital requirements.

The episode intensifies the broader debate in Washington over the influx of Chinese open models. The argument is that restrictions or bans are needed to preserve America’s technological advantage and mitigate potential national security risks.