US Treasury Secretary Scott Bessent has warned Kimi K3 AI model maker Moonshot AI and other Chinese AI companies of ‘Entity List’ designations after accusing them of using outputs from American frontier models, and saying that the officials have identified ‘evidence of US Al models appearing in some Chinese systems’.“We support open-source AI and the innovation it unlocks. But open source is not open season on American IP. When PRC firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations will be on the table,” Bessent said in a post on X.US Entity List is a trade restriction tool that consists of groups that the US thinks are acting against the country’s national security. It also restricts sensitive technology exports and requires special government licenses.
‘Moonshot AI used Nvidia chips to train models’
Bessent’s comments came soon after a senior White House official has accused Chinese artificial intelligence (AI) startup Moonshot AI of illegally tapping into restricted Nvidia chips to train its new AI model that the company claimed is as powerful as Anthropic’s Fable 5. The allegations follow Moonshot’s release of Kimi K3, a massive open-source model with 2.8 trillion parameters. The model has sent shockwaves through Silicon Valley after closing the performance gap with top American AI systems.Michael Kratsios, Director of the White House Office of Science and Technology Policy, publicly raised the alarm, accusing the Chinese company of circumventing strict US export controls to access Nvidia’s high-end hardware abroad.
Read full post by Michael Kratsios on X
We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model. To do this they developed a sophisticated internal platform to conduct large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection. Moonshot AI has also acquired GB300-equipped servers and has accessed GB300s in Thailand, likely to train its AI models.The United States strongly supports the free and fair development of AI, including a thriving competitive ecosystem that spans frontier models, specialized systems, open-source frameworks, and open-weight models. Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.Under current US trade restrictions, Nvidia’s most advanced AI semiconductors are strictly banned from being exported to China. While the US has permitted the use and sale of lower-powered chips, the high-performance Nvidia GB300 systems fall under tight restrictions due to their importance in training large-scale neural networks.
