Inside the White House debate over what to do about Chinese AI
The Trump administration still searching for a coherent China AI strategy as top officials pursue competing approaches.
By Sophia Cai
07/23/2026 07:08 PM EDT
The Trump administration is wrestling with how to respond to the possibility that leading Chinese tech firm Moonshot AI stole American intellectual property to create its latest artificial intelligence model.
The company’s Kimi K3 model – released last week – is nearly as powerful as the U.S.’s most advanced efforts but far less costly.
That creates a confounding dynamic inside the administration where President Donald Trump’s science and technology adviser and Treasury Secretary Scott Bessent are taking a hardline approach aimed at protecting American IP, while Commerce Secretary Howard Lutnick has a more modest approach, concerned with companies’ access to cheaper, effective technology.
The debate inside the White House underscores the competitive uncertainty and potential threats that Chinese firms represent to American tech companies and the degree to which the Trump administration has failed to coalesce around a response.
In the six days since Moonshot AI released its Kimi K3 model, Bessent and Office of Science and Technology Policy Director Michael Kratsios offered diverging policy ideas, none of which so far have been publicly announced by the White House.
“These conversations are taking place,” White House Press Secretary Karoline Leavitt said in response to a question from West Wing Playbook.
Axios reporting that the administration was considering banning Chinese open-weight AI models caused widespread panic in Silicon Valley, but no such sweeping proposal was seriously on the table at that point, according to two people familiar with the discussions, granted anonymity to discuss private conversations.
Bessent on Tuesday said on Fox Business’ “Mornings with Maria ” that the administration would investigate whether Chinese AI companies had improperly distilled American models to build their own systems and threatened sanctions.
“We have the ability to sanction them,” Bessent said.Moonshot did not immediately respond to comment.
On Wednesday, Kratsios went further, saying on X the administration had information suggesting Moonshot AI distilled Anthropic’s Fable model while developing Kimi K3, alleging the Chinese company created “a sophisticated internal platform” to conduct large-scale distillation against American AI systems while attempting to evade detection. He also alleged Moonshot had acquired Nvidia GB300-equipped servers despite U.S. restrictions on their sale to Chinese entities, but he did not provide evidence to back up the claims.
But Kratsios’ assertions immediately drew skepticism not only from AI researchers and industry but also from people in and around the administration.
“The whole distillation component at its core doesn’t quite make sense,” said one person close to the White House.
The person questioned whether it would be technically possible to distill – or copy – an advanced frontier model in only a matter of days.
“Especially layering in that Kimi is benchmarking ahead of [Anthropic’s Fable] in a number of ways,” the person said. “You can’t just distill your way ahead of somebody.”
The X post also exposed internal coordination problems, as OSTP and the Commerce Department that actually possesses export-control authority were not operating in lockstep.
According to the two people familiar with the discussions, Kratsios’ public comments were not made in coordination with the Commerce Department, which oversees the Entity List, a list of companies subject to export controls and licensing restrictions.
“OSTP at times is in the room, but Michael [Kratsios] is not seen as an effective executor of policy,” said the person close to the White House. “So they bring him in because OSTP has a lot of staff, and there’s no reason to undermine him. But he’s not always in the room.”
Similarly, the person added: “Commerce and Howard are sometimes in the room, but not always in the room. Sometimes these silos start working on things and maybe even produce things without sign-off from the West Wing and without total collaboration across government.”
Abdul El-Sayed on AIPAC spending, ‘Defund the Police’ and why he’s not a socialist
“The President’s team is working closely together to balance innovation and security in AI policymaking,” White House spokesperson Liz Huston said in a statement.
The Commerce Department is in no rush to amend the Entities List. Commerce’s Center for AI Standards and Innovation worked with the UK Artificial Intelligence Security Institute on a joint evaluation of Moonshot AI’s latest model and found that the newest Chinese AI model is still behind U.S. models, but it’s the strongest open-weight model.
Commerce officials have not begun drafting plans to place Moonshot or other Chinese AI companies on the Entity List, according to a person familiar with the department’s thinking. Any decision, the person said, would depend on conclusive findings by the Commerce Department.
Bessent, meanwhile, sought to leave every option available.
“We support open source AI and the innovation it unlocks,” he wrote Wednesday on X. “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.”
His position reflects more than AI policy.
As the administration official expected to lead the next round of U.S.-China economic negotiations, Bessent is both concerned about Chinese AI and has incentives to preserve leverage over Beijing. Sanctioning major Chinese AI companies or placing them on the Entity List would reverberate well beyond the technology sector and could complicate broader trade talks expected later this year.
Meanwhile, the administration is facing competing pressures from the AI industry.
[@portabletext/vue] Unknown block type "most-read", specify a component for it in the `components.types` prop
Anthropic has welcomed increased scrutiny of Chinese AI developers, arguing they pose national security risks and have benefited from intellectual property theft. Any restrictions on Chinese competitors would also advantage American frontier model companies.
At the same time, nearly 200 startup founders this week urged the Trump administration not to cut off access to Chinese open-weight models, arguing such a move would disproportionately hurt smaller American AI companies that rely on inexpensive open models because they cannot afford the computing costs associated with frontier systems from Anthropic or OpenAI.
President Donald Trump has yet to publicly weigh in, leaving the debate largely to Cabinet officials and senior advisers still trying to determine where the administration ultimately wants to land.
“People don’t always know that they’re not the ones in the meeting making decisions,” the person close to the White House said. “There’s a difference between people having ideas and the administration deciding on policy.”
Loading
West Wing Playbook: Remaking Government
Your guide to Donald Trump's unprecedented overhaul of the federal government.
By signing up, you acknowledge and agree to our Privacy Policy and Terms of Service . You may unsubscribe at any time by following the directions at the bottom of the email or by contacting us here . This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Sign Up
\* All fields must be completed to subscribe
Read Original at Politico →