The Pentagon labelled Anthropic a "supply chain risk" in February 2026, cutting the company off from military contractors. On Thursday, in the Northern District of California, Judge Rita F. Lin heard argument on whether that designation survives review — and signalled unmistakably which way she is leaning.

What the judge said

Lin told government counsel she had "seen no evidence" supporting the assertion that Anthropic could disable or modify systems after delivery, or activate a kill switch during operations. On the government's reliance on the company's public criticism of Defense Department policy, she said: "I find that position really troubling and at odds with the First Amendment." She added that the government's case had "gotten worse" since the preliminary-injunction stage. DOJ counsel James Harlow argued that AI vendors warrant heightened caution because, unlike physical products, AI systems cannot be inspected.

What did not happen

This is where most of the coverage went wrong. The proceeding was oral argument on cross-motions for summary judgment. No ruling issued, no order was entered, and the docket shows nothing filed. Alternet ran "Judge torpedoes Trump admin plan in new ruling"; Crypto Briefing ran "Judge rules Trump administration lacks evidence". Neither is true. Courthouse News, reporting from the room, states explicitly that no final decision was issued and that Lin merely indicated she will likely rule for Anthropic. The timing of that decision is undetermined.

And it would not be a new win

A preliminary injunction has blocked the designation since March 2026. A ruling for Anthropic would convert existing relief into permanent relief, not create protection that does not exist today. The case is Anthropic PBC v. U.S. Department of War, No. 3:26-cv-01996, brought under the Administrative Procedure Act. A separate petition for review sits in the D.C. Circuit as No. 26-1162, so the question is moving on two tracks.

What is actually being tested

The dispute began when negotiations collapsed after Anthropic declined to allow its frontier models to be used for mass surveillance, targeting decisions or lethal-force systems. The government's theory is that this refusal, combined with public criticism, makes a vendor a national-security risk. Whichever way Lin rules will set the terms on which any lab can decline a military use case without forfeiting federal business.