US Treasury Secretary Scott Bessent said on 21 July 2026 that Washington would investigate open-source Chinese AI models for evidence of intellectual property theft, warning that sanctions could follow against Chinese firms found to have stolen from American companies. The move reflects growing US concern that Chinese models, such as Moonshot AI's recently released Kimi K3, are rapidly closing the capability gap with American systems, threatening the commercial position and fundraising prospects of leading US labs including OpenAI and Anthropic.

The threat follows a report that the Trump administration is considering an outright ban on Chinese open-source models, though this has been disputed, and builds on earlier White House pledges in April to work with AI firms to counter alleged IP theft. It would extend existing US efforts to curb China's AI progress, which have already included restricting access to advanced chips. Central to the dispute is "distillation", a technique for transferring capabilities from large models into smaller ones; opinions differ on whether this constitutes theft, with critics such as Microsoft's Satya Nadella and Hugging Face's Clem Delangue arguing the practice is widespread and only a minor factor behind China's progress, pointing instead to strong Chinese research teams.