Trump AI advisors traded insults over Kimi, a free China model that undercuts US frontier pricing

Over the weekend, several current and former advisors to President Donald Trump on AI publicly lobbed insults at the country’s leading AI companies, and the trigger has a surprisingly simple name: Kimi. The issue is what happens when China ships a free, open source model that appears to rival intelligence from OpenAI and Anthropic, at a time when US AI companies are heavily dependent on selling access to frontier models.

David Sacks, the president’s AI and crypto “czar” until March, accused Anthropic’s models of being “lobotomized” and “woke.” Emil Michael, described as a top Pentagon official, called OpenAI’s new head of strategic futures a “supreme village idiot.” That sparring looks personal. But the center of gravity is policy. Kimi and other Chinese models like it create a “real problem” for Trump, according to the framing in the story, because every new smart, free model from China reduces US companies’ incentive to spend big money to access models from Anthropic or OpenAI.

There is also a financial and political urgency behind the argument. The story notes that enthusiasm for major AI companies is driving an outsized share of economic growth, so China’s AI models create economic and political problems for the president. Anton Leicht, a fellow at the Carnegie Endowment, wrote on X that they are “a threat for an administration that really doesn’t want more economic bad news.” And the story adds that the models have already rattled US stocks.

If you are a decision-maker in the US AI ecosystem, this is the uncomfortable part: the market and the messaging are converging in a way that makes it harder for everyone to pretend the debate is only technical. The US conversation is happening while New York has just imposed the country’s first state ban on new data centers, which raises the pressure on compute demand and the speed at which companies can scale. At the same time, there is growing distrust of AI companies, and the story points out that many Americans might have little sympathy for OpenAI or Anthropic as cheaper competitors from abroad arrive. In other words, even if US companies want government help, they may not get broad public support for protecting their business models.

Sacks’ position highlights that split. On July 19, he criticized top AI companies for wanting the government to eliminate their open source competition. He also argued that Chinese AI models have become popular because they come with fewer restrictions on how people can use them, putting aside the built-in state censorship the story references. But Sacks is out of a formal role now. His more open-leaning view has been replaced inside the administration by a different instinct: government intervention.

The story describes that as coming from a national security framing. The thinking is that AI models have gotten strong enough to pose threats to national security, so the government must control how they are used. That position underpins the new White House review process aimed at vetting AI models’ security before they are released. Dean Ball, described as a former Trump AI advisor now working for OpenAI, criticized the process over the weekend as a “de facto licensing regime for frontier AI.” Ball predicted Trump might handle the Chinese open source problem with soft power, perhaps by making US companies afraid to use models like Kimi.

Emil Michael pushed back hard. Along with Secretary of Defense Pete Hegseth, the story says Michael has been a main Pentagon liaison with AI companies. Michael called Ball the AI industry’s “supreme village idiot” and bristled at the suggestion of quiet strong-arming rather than, in his words, going through “the democratic process not some Deep State scheme.” That disagreement is the real reason this is more than drama. It is a fork in the road over whether the US response should be structural and legal, or coercive and informal, and both camps see the Chinese open source availability as a threat.

What complicates everything is that the story says it leaves out the single question that would matter most for crafting a durable strategy: how a model like Kimi got so good in the first place. For much of the Biden administration, and even the beginning of Trump’s second administration, keeping China from getting top chips was a priority. Those export controls have loosened. The story notes that Trump made the controversial decision to allow Nvidia to sell more chips to China, in exchange for the US government taking a cut, and it says the government has alleged some chip smuggling has taken place. But even with loosening, China has limited computing power, and it is not clear what chips the company behind Kimi used to train the model.

One possibility mentioned is distillation, where AI models are trained on the outputs of existing AI models. The story says OpenAI and Anthropic have long complained that Chinese companies do this, and they requested government help to stop it. In April, the Trump administration announced efforts to curb distillation. Still, Kimi is out there and free, and the story emphasizes that it is nearly as good as the Anthropic model the US government deemed so powerful that it was briefly shut down because it threatened national security.

So where does this leave Trump’s orbit? The weekend’s sparring suggests many in the administration see a wake-up call. But nobody can agree on what for. For US AI companies, the stakes are straightforward: new free competition can reduce willingness to pay for access to frontier models, right when their growth narratives depend on it. For government actors, the stakes are tougher: tighter security vetting and licensing-like regimes could slow the release of models, but they might also fail to stop the flow of strong open models from overseas.

And for executives, the second-order effect is that the public fight is training the market to expect policy whiplash. If the administration can argue openly about insults, licensing regimes, and soft power in the same breath, then procurement, partnerships, and go-to-market plans become harder. The board-level question becomes: are we building for the world where access is controlled, or the world where competitive pressure arrives from open, free models outside the fence? Either way, Kimi turned a technical release into a pressure test for how the US state and US markets coordinate under uncertainty.