The release of Kimi K3, a free open-source AI model from Chinese company Moonshot AI, has exposed sharp divisions among the top artificial intelligence strategists advising President Donald Trump.
Last weekend, several current and former administration advisors publicly attacked U.S. AI companies while arguing over how to respond. David Sacks, the administration's AI and crypto czar, branded Anthropic's models "lobotomized" and "woke." Emil Michael, a top Pentagon official, called OpenAI's new head of strategic futures a "supreme village idiot." The insults were not random. They were symptoms of a deeper disagreement about what to do when a capable Chinese model is released at no cost.
The economic logic is straightforward. Every time a powerful, free open-source model emerges from China, the commercial case for paying U.S. providers like OpenAI and Anthropic weakens. For the Trump administration, which has made American AI dominance a priority, this creates a policy problem with no clean solution. Restricting open-source models is technically difficult and politically fraught. Ignoring them risks ceding influence over the global AI stack.
The debate is not just about one model. Kimi K3 appears to rival the intelligence of leading closed models from U.S. labs, according to early assessments. If that holds up under broader testing, it would mean the capability gap between American and Chinese frontier models has narrowed further, and that open-source release strategies can nullify some of the spending advantage that U.S. companies have built through massive capital investment.
The infighting among Trump's advisors reflects a broader uncertainty in Washington about how to handle AI competition with China when the technology moves faster than policy. Traditional tools, export controls on chips, investment restrictions, tariffs, were designed for physical goods and concentrated supply chains. Open-source software distributed globally over the internet does not fit that framework well.
What comes next depends on whether the administration can coalesce around a coherent approach or whether the current public fractures harden into competing factions with different visions for American AI strategy. The Kimi K3 episode suggests that cohesion is still a long way off.