Judge Rita F. Lin of the Northern District of California issued a 43-page ruling on Thursday granting Anthropic a preliminary injunction against the Pentagon’s decision to designate the company a “supply chain risk” and against a Trump executive order barring federal agencies from using Anthropic’s technology. The language in the opinion was not subtle. “Nothing in the governing statute,” Lin wrote, “supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government.”
She went further. The Pentagon’s actions, Lin found, constituted “classic illegal First Amendment retaliation.” The government punished Anthropic not for any supply chain failure, not for any security breach, not for any deficiency in its technology, but for CEO Dario Amodei’s public statements that Claude would not be used for autonomous weapons systems or to surveil American citizens without consent.
This matters because the case just jumped categories. What started in February as a Pentagon procurement dispute over AI safety guardrails, escalated in March into a lawsuit and a hearing where Silicon Valley split publicly over which side to back, and has now landed in constitutional law territory. A federal judge looked at the evidence and concluded that the United States government retaliated against an American company for exercising free speech.
That is not a procurement ruling. That is a First Amendment precedent.
The injunction temporarily blocks the supply chain risk designation and the federal agency ban, but Lin put her own ruling on hold for one week to give the Justice Department time to file an appeal. The DOJ is expected to take it. Which means the Ninth Circuit will likely be weighing in on whether the executive branch can use supply chain designations as political punishment within the next few months, and that ruling will set precedent that reaches far beyond one AI company.
Why this outlasts the current administration
The specific fight between Anthropic and the Pentagon will resolve one way or another. Either the appeals court upholds Lin’s injunction and the designation gets scrapped, or it reverses and Anthropic faces years of litigation while locked out of federal contracts. But the constitutional question Lin raised, whether the government can use procurement and supply chain authorities to punish companies for public speech, applies to every defense contractor, every technology vendor, and every company that does business with the federal government.
David Sacks left the White House AI role with no replacement named. The administration’s AI policy framework is a preemption play designed to block state regulation rather than articulate any actual federal standard. And now a federal judge has called the executive branch’s treatment of its most prominent AI safety advocate “Orwellian.” The administration’s AI posture is becoming incoherent in a way that the judiciary is starting to notice.
The counter argument
The injunction is preliminary, not final. Lin’s ruling survives only if the Ninth Circuit agrees, and appellate courts regularly narrow or reverse district court injunctions, particularly in national security cases where courts traditionally defer to executive branch discretion. Politico reported that lawyers and lobbyists familiar with the case say Anthropic is “still in trouble” despite the ruling, because the underlying procurement dispute has not been resolved and the political dynamics have not changed. The government may lose the constitutional argument and still find administrative ways to exclude Anthropic from contracts.
This is not about procurement anymore and it has not been for weeks. A federal judge just told the Pentagon it cannot punish a company for disagreeing with the government’s position on how AI should be deployed in military contexts. That principle, that the First Amendment applies even when the disagreement is with the Department of Defense during wartime, is a constitutional marker that will outlast this administration, this case, and probably this generation of AI technology. The specific contract dollars at stake are a rounding error compared to the precedent being set.
What This Means for Everyday People
The AI tools you use every day exist because companies like Anthropic make decisions about what those tools will and will not do. When a government can punish a company for choosing to put safety limits on its technology, the incentive for every other AI company is to remove those limits to avoid the same treatment. This ruling protects not just Anthropic’s right to speak, but the idea that the companies building AI get to have an opinion about how it should be used without being blacklisted for saying so.