The Trump administration is divided over how to deal with the rise of Chinese leaders Examples of AIaccording to many people familiar with the discussions.
The debate is broadly divided between parts of the White House, which has pushed for tighter controls on Chinese AI that could soon rival more powerful US models, and the Commerce Department, which has seen such restrictions as unenforceable.
The conundrum has become the latest benchmark for AI dominance, after China’s Moonshot AI lab last week released its Kimi K3 model that competes with top models from the US. Anthropic and OpenAI.
In recent days, senior White House officials have discussed possible steps to prevent Chinese AI labs from building their own models by teaching them on American models, a practice known as distillation, the people say.
The White House is expected to take some form of presidential action after last month Anthropic sued the Chinese technology company Alibaba he had made “the biggest distillation attack” against him to date. The move is not currently expected to take the form of an executive order, according to one of the people.
A White House spokesman had no comment on the reporting in this story.
The White House has been taking attacks on distillation particularly seriously. It is it was announced on Wednesday that Moonshot had created Kimi K3 by manipulating the Anthropic design of Claude Fable 5, and Treasury secretary Scott Bessent. he told Fox Business that secret distillation can lead to restrictions, and describe it as “IP theft.”
While the proposals have been making the rounds within the White House, no formal request for input has been sent to the Commerce Department, which has its own views, according to two people familiar with the deliberations.
The Commerce Department has increasingly emerged as an important part of the domestic regulatory divide over AI, in part because it oversees export controls through the Office of Industry and Security.
Lutnick has considered ways to create incentives for major U.S. labs to develop their own open-weight models to counter China, and has spoken with leaders at several AI labs in recent weeks, the people add.
Lutnick appears to have moved around a central core of control—he put shipping controls on Anthropic to bring them back—but has been more independent than Cairncross, one of the people said.
Earlier this month, the biggest AI issue for the administration was Claude’s Anthropic Structures Mythos and Fable 5because of their ability to find weaknesses in government systems and critical infrastructure.
But policy and AI headaches have only increased with the rise of so-called open-source models from China—models that have no protections against government hacking and can be downloaded and used by anyone.
This is a version of Picture of Hugo Lowell Inner Loop Journal. Read previous articles here.
