US Treasury Secretary Scott Bessent repeated his warning to Chinese artificial intelligence companies on Wednesday, saying the United States could impose sanctions over alleged intellectual property theft.
His comments followed an accusation from a White House official that China-based Moonshot had improperly used Anthropic’s Fable model to train its own AI system through model distillation.
Model distillation is a common training method in which a smaller AI model learns from the outputs of a larger model. The technique can violate intellectual property rights in some cases, but companies also widely use it as a legitimate way to optimize AI models.
“Open source is not open season on American IP,” Bessent wrote on X.
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.
— Treasury Secretary Scott Bessent (@SecScottBessent) July 22, 2026
He said sanctions and additions to the US Entity List would remain possible when Chinese companies conduct secret, industrial-scale distillation operations that cross into intellectual property theft.
Earlier this week, Bessent said the US government would examine open-source models developed in China for signs of stolen intellectual property. He said authorities would impose sanctions if they found evidence of theft.
Bessent issued his latest warning hours after Michael Kratsios, the White House’s science and technology policy chief, accused Moonshot of conducting large-scale distillation using American AI models.
Kratsios also alleged that Moonshot had obtained servers equipped with Nvidia’s GB300 chips and accessed other GB300 systems in Thailand, likely to train its AI models.
The allegation raised questions about whether Moonshot had violated US export-control rules.
The GB300 servers belong to Nvidia’s Blackwell generation. US restrictions prohibit companies from selling these systems to Chinese firms.
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…
— Director Michael Kratsios (@mkratsios47) July 22, 2026
Some experts dispute the claim that Moonshot primarily developed its Kimi K3 model by distilling Anthropic’s Fable model.
Fable has only been publicly available since July 1, while Moonshot released K3 last week as an open-weight model.
K3’s advanced capabilities have raised questions about the business models of leading US AI laboratories. They have also created doubts about whether those companies can continue to justify the enormous capital spending required to compete in the frontier AI race.
The dispute has intensified a broader debate in Washington over the growing number of open AI models coming from China.
Some officials and industry figures have called for tighter restrictions on these models.
Dean Ball, a former White House AI adviser who now serves as OpenAI’s head of strategic futures, has argued that the United States should restrict or effectively ban Chinese open-weight models.
He said such measures could help protect America’s technological advantage and reduce possible national security risks.
Get the latest tech news, telecom insights, and product launches wherever you prefer.
Add ProPakistani to Preferred Sources and see more of our stories in Google Search and Top Stories.
Shares
