Trump Says He Fired Anthropic ‘Like Dogs’

President Donald Trump took credit for decisive action against Anthropic on Thursday, as the AI company still waits to be formally designated as a “supply chain risk” by the Pentagon. And he used one of his favorite phrases in the process.

“Well, I fired Anthropic. Anthropic is in trouble because I fired [them] like dogs, because they shouldn’t have done that,” Trump said in a new interview with Politico.

Trump has a long history of saying that such-and-such person died like a dog (ISIS leader Abu Bakr al-Baghdadi) or lied like a dog (Ted Cruz) or “cheated on him like a dog” (Kristen Stewart). And while it often makes no sense, that’s typically his definitive statement on the matter. The question is whether we should read much into it this time.

The tech world is waiting to see how the Pentagon’s battle with Anthropic shakes out, given the stakes for the future of AI. The Department of Defense had demanded that Anthropic drop guardrails on its AI model Claude that prohibit its use in mass domestic surveillance and fully autonomous weapons. That didn’t go over well with Defense Secretary Pete Hegseth, who wrote on X that he was designating Anthropic as a supply chain risk.

That designation, at least according to Hegseth, means that other businesses that work with the U.S. government can’t work with Anthropic, a move that’s been described as a form of corporate murder. Anthropic instantly promised to file suit if the Pentagon was going to do something so drastic. But Anthropic has yet to acknowledge being given formal notice, leaving everything in a state of limbo.

Bloomberg reported Thursday that it had given the company notice, but the source is an anonymous official who “didn’t say when or by what means the Pentagon informed Anthropic of the designation.”

The latest reporting from the Financial Times and Bloomberg indicates Anthropic CEO Dario Amodei is currently in talks with the Pentagon, suggesting that there could still be some form of agreement that’s reached in the near future.

The discussions are being held with Emil Michael, Under-Secretary of Defense for Research and Engineering, a man who last week tweeted that Amodei was a “liar” with a “god-complex.” So it remains to be seen how that will go, especially since it seems unlikely Anthropic can compromise after making such a big deal about ethics.

On the other hand, Anthropic’s corporate survival is on the line. And it’s not like the company is some bastion of moral behavior right now. The U.S. military is currently using Claude to help choose targets in the bombing of Iran, according to the Wall Street Journal. Specifically, the military is using Claude as it’s embedded into Palantir’s Maven Smart System, according to the Washington Post.

From the Post:

As planning for a potential strike in Iran was underway, Maven, powered by Claude, suggested hundreds of targets, issued precise location coordinates, and prioritized those targets according to importance, said two of the people. The pairing of Maven and Claude has created a tool that is speeding the pace of the campaign, reducing Iran’s ability to counterstrike and turning weeks-long battle planning into real-time operations, said one of the people. The AI tools also evaluate a strike after it is initiated, the person said.

It’s unclear whether the U.S. or Israel was the country that bombed a school in Minab, Iran, that killed 168 people, most of them children.  But there’s speculation that it was the U.S. (Minab is in the south, territory the U.S. is primarily bombing) and that they could have been using Claude.

The school was built on an old Revolutionary Guard base that closed about 15 years ago, according to NBC News. Did the U.S. military use AI to choose the target, and did it have old information that it failed to properly check? Plenty of people are talking about that possibility right now, and it all seems too plausible.