Chinese model Kimi splits Trump’s AI circle

Kimi, a free, open source model launched last week by Chinese AI company Moonshot, has intensified a dispute inside President Donald Trump’s AI orbit. The model appears to rival systems from OpenAI and Anthropic, which are not free, creating a competitive threat at a time when enthusiasm for US AI companies is helping drive economic growth.

Several current and former Trump advisers publicly clashed over how the government should respond. David Sacks criticized leading AI companies on July 19 for wanting the government to eliminate open source competition, and has argued that Chinese models are gaining traction because they impose fewer usage restrictions, aside from state censorship. Others around Trump favor stronger intervention, arguing that powerful models now raise national security concerns.

The White House has created a review process to assess model security before release, drawing criticism from Dean Ball, a former Trump AI adviser now at OpenAI, who called it a “de facto licensing regime for frontier AI.” Pentagon official Emil Michael rejected suggestions that the government might quietly pressure companies not to use models like Kimi, saying policy should go through “the democratic process not some Deep State scheme.”

Kimi’s rise also raises questions about chips, export controls and distillation, a method in which models train on outputs from other systems. OpenAI and Anthropic have sought government help against the practice, and the Trump administration announced efforts to curb it in April.