Key facts
- White House official Michael Kratsios accused China's Moonshot AI of distilling Anthropic's Fable model to develop Kimi K3.
- Moonshot AI allegedly developed an internal platform to distill U.S. models at scale, using methods designed to evade detection.
- U.S. Treasury Secretary Scott Bessent stated sanctions and Entity List designations are possible for Chinese firms conducting covert, industrial-scale AI distillation attacks.
- Some AI researchers question the technical feasibility of Kimi K3's performance being derived from distillation of Anthropic's Fable model.
- The U.S. government is concerned about Chinese firms accelerating AI progress through alleged intellectual property theft.
The U.S. Treasury is considering sanctions against Chinese artificial intelligence companies for alleged intellectual property theft, following accusations from a White House official that China-based Moonshot AI improperly distilled Anthropic's Fable model for its K3 release. Treasury Secretary Scott Bessent warned that sanctions and Entity List designations would be considered for such actions, emphasizing that 'open source is not open season on American IP.'
White House Office of Science and Technology Policy Director Michael Kratsios alleged that Moonshot AI conducted large-scale, covert industrial distillation attacks against U.S. models, using methods designed to evade detection. He also suggested the firm may have acquired restricted Nvidia GB300 servers and accessed them in Thailand, potentially violating U.S. export control rules.
Model distillation, a technique where a smaller AI model learns from a larger one, is a common optimization method but can infringe on intellectual property rights. Some experts dispute the claim that Moonshot AI's Kimi K3, released on July 16 as an open-weight model, could have been primarily developed through distillation from Anthropic's Fable model, which was only publicly available since July 1. Elie Bakouch, a researcher at AI startup Prime Intellect, noted the narrow 15-day window between Fable 5's ban removal and Kimi K3's release, questioning the technical feasibility of such an attack. Dean Ball, OpenAI’s head of strategic futures, also expressed skepticism.
The incident has intensified a broader debate in Washington regarding the influx of Chinese open-weight models, with some advocating for restrictions to preserve U.S. technological advantage and mitigate national security risks.