Trump faces pressure over Chinese AI models and possible US restrictions

US President Donald Trump is under pressure over the growing success of Chinese artificial intelligence models that many companies prefer because they are cheaper than more powerful US systems from companies such as Anthropic and OpenAI.

The debate intensified after Moonshot released Kimi K3. White House officials said the model was developed by taking capabilities from Anthropic's most powerful model through a process called distillation. The method is widely used in the AI industry, but US officials say Chinese companies have used it at scale to copy proprietary American systems illegally.

Michael Kratsios, director of the White House Office of Science and Technology Policy, criticised the practice on Wednesday. He also suggested that Moonshot had avoided US export controls on powerful Nvidia AI chips while building its products.

A spokesperson told The Information that the Commerce Department's Bureau of Industry and Security is formally investigating Chinese companies including Moonshot over their use of those chips. Treasury Secretary Scott Bessent also threatened sanctions against China this week, as reports said the United States was considering a ban or limits on foreign-made open source models that are often built through distillation.

Earlier this year, Anthropic sent letters to US lawmakers warning about the practice. The company accused Moonshot, DeepSeek and MiniMax of industrial-scale distillation of its Claude models. Sarah Heck, Anthropic's head of public policy, said on X that illicit distillation was intellectual property theft and industrial espionage, and that it created national security risks for the United States and its democratic allies.

Critics of a tougher policy say limits on distillation could also become a ban on open source, or open-weight, AI models. These models cost less and can be customised by programmers, making them useful for startups and companies that do not want to rely on paid proprietary systems from Anthropic, OpenAI or Google.

The idea of blocking foreign-made open source models has drawn opposition from parts of Silicon Valley. Opponents say such a move would benefit large US AI companies, reduce competition and leave entrepreneurs paying higher prices for AI tools.

On Wednesday, 179 startups wrote to the Trump administration urging it not to impose an outright ban or strict curbs on foreign-made models, except possibly for government use. The Little Tech Association said denying US startups access to models available abroad would hurt competition, strengthen incumbent companies and act as a tax on intelligence.

On Friday, Microsoft, Mistral, Perplexity and Meta also called for restraint, including in cases involving illegal distillation. They proposed dealing with the issue through legal and commercial rules rather than broad restrictions.

Some critics say OpenAI and Anthropic are seeking government regulation to slow rival technology. The two companies are under scrutiny ahead of expected initial public offerings in coming months, with investors assessing their business performance.

David Sacks, the former White House AI policy chief who still has the president's ear, said on X that the United States would continue to lead if it did not undermine itself with unnecessary rules.

Nvidia chief executive Jensen Huang also argued in favour of allowing use of Chinese models. He told Axios that American companies should be allowed to use Chinese AI models, and said strong open source models should be used.