The Trump administration has dropped plans to sanction Chinese open-source AI models like Kimi K3, following strong opposition from major tech firms.
According to the New York Times, officials including White House Chief of Staff Susie Wiles, Treasury Secretary Scott Bessent, and Commerce Secretary Howard Lutnick debated aggressive intervention in the open-source sector. The group considered trade blacklists, sanctions, and restrictions preventing US cloud companies from working with Chinese developers. They also weighed rules for federal agency access to these models.
Five sources told the paper that the administration is now shifting focus to improving the competitiveness of American models rather than imposing hard measures.
The debate intensified after Moonshot AI released Kimi K3, a system matching top US performance in several areas. Michael Kratsios, head of the White House Office of Science and Technology Policy, accused the firm of distilling Anthropic‘s model Fable on X. He called the alleged theft of US technology unacceptable. Bessent floated sanctions that same day.
A business fight dressed up as national security
OpenAI and Anthropic pushed for the restrictions, citing national security. Both firms keep their models closed. Executives at Nvidia, Microsoft, Google, and Meta feared the two companies could use the White House to lock in their market position, especially as they plan initial public offerings.
Through private messages and video calls, the tech giants built a counter-argument. They claimed open models speed up innovation and aid cyber defence. Nvidia CEO Jensen Huang became the face of this resistance. On July 24, he posted his first-ever message on X defending open models.
Nvidia also launched an alliance for open security tools. It grew from a handful of members to more than 230. Christopher Padilla, a former Commerce Department official, described the dispute in the New York Times as a cage fight over the future of the AI industry. He noted the conflict is mostly about money and market power, with security concerns added on. Industry officials do not expect hard measures before President Xi Jinping’s visit in September.
OpenAI told the paper that both open and closed models have a role to play. The company argued that progress in Chinese open-weight models is not a case against openness but a reason for a national framework. Anthropic declined to comment, though CEO Dario Amodei has previously outlined his position on open source.
The New York Times has a pending copyright lawsuit against OpenAI and Microsoft.
