Judge Shoots Down Trump Admin’s Attempt to Blacklist Anthropic

Well, that didn’t take long to fall apart.

A federal judge just handed the Trump administration a pretty clear loss in its attempt to blacklist Anthropic. US District Judge Rita Lin didn’t mince words when she called the Department of War’s effort to designate the AI company a supply-chain risk “classic First Amendment retaliation.”

Her order granting Anthropic’s request for a preliminary injunction is worth reading if you care about how far government overreach can go before someone slaps it down. Lin wrote that, by all appearances, “these measures appear designed to punish Anthropic.” That’s not the kind of language judges use when they think the government has a solid case.

The core problem here is straightforward: officials apparently had no real authority to take such extreme actions. They didn’t consider less restrictive alternatives. They didn’t offer any evidence that Anthropic posed an urgent risk to national security. Instead, the Department of War’s own records showed they designated Anthropic as a supply chain risk because of its “hostile manner through the press.”

Let that sink in. The government tried to blacklist a major AI company because it was being difficult in media interviews.

I’ve seen a lot of weak justifications for regulatory action over the years, but this one is particularly embarrassing. It’s not even pretending to be about security. It’s about hurt feelings and a company that wouldn’t play ball publicly.

What’s especially striking is how quickly the legal system stepped in. Preliminary injunctions aren’t handed out like candy. The court has to believe there’s a substantial likelihood of success on the merits, and that irreparable harm will occur without intervention. Lin clearly saw both.

This is higher than I expected in terms of judicial pushback. I figured we might see some procedural delays or requests for more briefing. Instead, we got a clean, direct ruling that calls out the administration’s play for what it is.

The First Amendment angle is the real kicker. Retaliating against a company for exercising free speech rights is about as blatant a constitutional violation as you can get in this context. The government can’t just decide to cut off a company’s access to contracts and markets because it doesn’t like what executives say in interviews.

Anthropic’s legal team did their homework here. They framed this not as a technical procurement dispute but as a fundamental rights issue, and the judge bought it.

Now, I’m not naive about national security concerns around AI. There are legitimate questions about how these models get deployed, who has access, and what safeguards exist. But none of that was actually at issue here. The government didn’t present evidence of security risks. They presented evidence of a company being difficult in the press.

That’s not how supply chain risk designations are supposed to work. Those tools exist for real threats, not for settling scores with companies that annoy political appointees.

Where this goes from here is anyone’s guess. The preliminary injunction keeps things frozen while the case proceeds. The administration could try to build a real evidentiary record, though that would require actually finding evidence that Anthropic poses a security risk. Or they could drop the whole thing and save face, which is probably the smart play.

But this ruling sends a signal to other agencies thinking about similar tactics. If you’re going to target an AI company for political reasons, you better have something more than “they were mean to us in the press.” The courts are watching.