Judge: No Proof Anthropic Poses Supply-Chain Risk, Rebukes Pentagon's Retaliation Logic

A federal judge delivered a sharp rebuke to the Department of Defense on Thursday, signaling the Trump administration’s case for blacklisting AI firm Anthropic is built on shaky ground. U.S. District Judge Rita Lin said the government hasn’t mustered enough evidence to justify labeling the company a supply-chain risk, a designation that would bar federal agencies from using its Claude models. The hearing is the latest flashpoint in a bitter dispute that began when Anthropic resisted letting the Pentagon use its AI for mass surveillance or lethal targeting, arguing the technology simply isn’t ready for that kind of deployment. The DOD shot back with a familiar refrain — private companies don’t get to dictate military use — and then made a move Lin found deeply troubling. The government claimed Anthropic’s public criticism of the DOD was itself grounds for the ban.

That logic stopped Lin cold. During the hearing, she warned it could establish a dangerous precedent where the government retaliates against any federal contractor who dares to disagree with the administration in public. It’s an argument that sounds more like punishing speech than managing risk. The DOD didn’t stop there, floating a hypothetical that Anthropic could potentially disable or alter its AI models during active warfighting operations — essentially flipping a kill switch on the battlefield. Experts and the judge were unimpressed. Lin said she saw no proof whatsoever that Anthropic could remotely modify a model after delivery, calling the scenario entirely unsupported.

The hearing is part of one of two lawsuits Anthropic filed in March challenging the ban and the supply-chain risk label. Lin had already temporarily blocked the ban that month, and she’s now weighing whether to make that order permanent. The other case is being heard in a Washington court. For Anthropic, the stakes are existential — being shut out of federal contracts is a massive commercial blow. For the DOD, the case tests how far it can go in strong-arming AI companies that want to sell to the government but set ethical boundaries on how their tools are used.

What makes this messy isn’t just the legal arguments. It’s the collision of two fundamentally different views of AI in warfare. Anthropic is drawing a line in the sand on autonomous lethal decisions, betting that a cautious stance now will look prescient later. The Pentagon sees that as an overreach by a vendor. But the judge’s skepticism of the kill-switch theory and the retaliation-by-designation play suggests the government’s legal strategy may need more than hypotheticals to stick. The question now is whether Lin makes her temporary block permanent — and whether that sends a signal to every other AI lab watching this fight unfold.