Updated
Updated · MIT Technology Review · Jul 20
Trump AI Advisers Split Over Free Chinese Model Kimi as US Firms Face New Open-Source Threat
Updated
Updated · MIT Technology Review · Jul 20

Trump AI Advisers Split Over Free Chinese Model Kimi as US Firms Face New Open-Source Threat

3 articles · Updated · MIT Technology Review · Jul 20

Summary

  • Kimi’s launch last week has exposed a sharp split inside Trump’s AI orbit, with current and former advisers publicly attacking OpenAI, Anthropic and each other over how Washington should respond.
  • The free, open-source Chinese model appears close to top US systems, undercutting paid offerings from OpenAI and Anthropic and creating an economic and political problem for an administration already wary of more bad market news.
  • David Sacks argued US companies want government protection from open-source rivals, while former adviser Dean Ball attacked the White House’s new security review as a de facto licensing regime for frontier AI.
  • Emil Michael, the Pentagon’s main AI liaison with Pete Hegseth, rejected Ball’s suggestion of quiet pressure on companies and said any response to Kimi should go through the democratic process.
  • The clash also revives questions about US chip policy and Chinese training methods: Trump loosened Nvidia sales to China, Washington has alleged chip smuggling, and OpenAI and Anthropic have pushed for action against model distillation.

Insights

As China dominates open-source AI, is the US debate on protecting its closed models a losing battle?
If a free Chinese AI becomes the world's standard, what are the hidden costs for global security and innovation?
With China's AI 'stealing' US tech, can new laws stop a brain drain that chip bans could not?