A Biden-appointed federal judge struck down the Pentagon's ban on AI company Anthropic, ruling that Defense Secretary Pete Hegseth's "supply chain risk" designation was illegal retaliation for the company's public criticism of military AI policy.

U.S. District Judge Rita Lin issued a 43-page ruling Thursday voiding the Pentagon's blacklisting of Anthropic, the maker of the AI system Claude. Lin found that the Defense Department branded Anthropic a national security threat not because the company posed any genuine danger, but because it had the nerve to push back, publicly, against how the military wanted to use its technology. The Daily Caller News Foundation reported that neither the Pentagon nor the Department of Justice responded to requests for comment on the ruling.

The clash traces back to February, when Anthropic refused to let the Pentagon use Claude for fully autonomous weapons systems and mass surveillance of American citizens. Hegseth had threatened the company with termination of its $200 million contract or a supply chain risk designation if it didn't open Claude up for all lawful military purposes. When Anthropic held firm, the Pentagon pulled the trigger on the designation, and President Trump ordered every federal agency to stop using Anthropic's products immediately.

Lin called the Pentagon's "supply chain risk" label an Orwellian punishment for free speech

The judge did not hold back. In her ruling, Lin wrote that the government's actions against Anthropic had nothing to do with legitimate security and everything to do with making an example of a company that spoke out.

As AP News reported, Lin found the government's conduct was "based on a desire to make a public example out of Anthropic for its arrogance in criticizing the government, not based on any articulable basis to believe that Anthropic would actually sabotage its model."

She went further. The government's own behavior after slapping the designation on Anthropic, Lin wrote, was not "consistent with a genuine fear that Anthropic is a saboteur who would poison its software to harm national security." If the Pentagon truly believed Anthropic was a threat, it had a strange way of showing it.

Lin also took direct aim at the legal theory behind the ban. The Washington Examiner reported that she called the designation "Orwellian," writing:

"Nothing in the governing statute 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."

That language matters. Lin framed the Pentagon's action as classic First Amendment retaliation, punishing a private company for saying things the government didn't want to hear. "Punishing Anthropic for bringing public scrutiny to the government's contracting position is classic illegal First Amendment retaliation," she wrote.

Hegseth, who has made a series of consequential personnel decisions since taking over the Pentagon, including removing top commanders in Europe, described Anthropic CEO Dario Amodei and his company as a "master class in arrogance" and a "textbook case of how not to do business with the United States Government." He said the government could not allow the military to be constrained by a private vendor's policy preferences.

Anthropic refused to remove safety guardrails, and the Pentagon retaliated

The dispute began when the Pentagon sought broad access to Claude, the only commercial AI system approved for classified military use at the time. Anthropic pushed back, citing reliability concerns and refusing to strip internal guardrails that prevented the system from being used for autonomous weapons targeting or domestic mass surveillance.

Hegseth personally labeled Anthropic a supply chain risk in February after the company refused to comply. President Trump amplified the fight on Truth Social, posting: "THE UNITED STATES OF AMERICA WILL NEVER ALLOW A RADICAL LEFT, WOKE COMPANY TO DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS!" He added: "We don't need it, we don't want it, and will [sic] not do business with them again!"

The administration's position had a reasonable core: the military should not be held hostage by a contractor's political preferences. If the Pentagon pays for a tool, it should be able to use that tool for lawful purposes. Hegseth has shown throughout his tenure that he is willing to make bold, sometimes controversial decisions, from stripping security clearances to restructuring how the department operates.

But Lin's ruling drew a sharp line between choosing not to do business with a company and punishing one for speaking up. The Pentagon was "undisputedly free to select the AI vendor of its choice," Lin wrote. What it could not do was weaponize a supply chain risk designation, a tool designed to flag genuine sabotage threats, as a cudgel against public criticism.

Despite the feud, the military kept using Claude in combat

One of the more striking details to emerge from the case is that even as the Pentagon publicly branded Anthropic a security risk, the military continued relying on Claude in the field. Fox News reported that the U.S. military used Anthropic's Claude AI to strike over 1,000 targets in the first 24 hours of Operation Epic Fury against Iran.

That fact cuts both ways. It shows how deeply embedded Claude had become in military operations, and it undercuts the Pentagon's claim that Anthropic posed a genuine threat to the supply chain. If the company were truly a saboteur capable of poisoning its own software, the military had an odd way of hedging that risk: by using the software to run a major combat operation.

Anthropic, for its part, struck a conciliatory tone after the ruling. A spokesperson told the Daily Caller News Foundation: "We welcome the court's ruling that this supply chain risk designation was unlawful. We remain focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology."

Anthropic's lawyer, Michael Mongan, was more pointed, arguing that the Pentagon's actions "profoundly harm Anthropic" and "threaten more broadly to chill speech and debate on a very important issue." Lin agreed, calling the government's position "really troubling" and "at odds to me with the First Amendment."

An appeal looms, and a second case is already pending

The government is expected to appeal Lin's ruling. And the fight between Anthropic and the Pentagon is far from over. A separate case remains pending in the D.C. Circuit, where Anthropic is challenging another Pentagon supply-chain label, according to the New York Times. The Trump administration had previously lifted export controls on two of Anthropic's AI models, Fable 5 and Mythos 5, alongside an agreement for the company to cooperate closely with the government.

That earlier deal suggested the two sides could find common ground. The supply chain risk designation blew it up. Hegseth, who has not shied away from reshaping the Pentagon's leadership and policies, including overhauling the department's recognized faiths list, now faces a federal court telling him he went too far.

The ruling raises a question that will outlast this particular case: where does the government's authority to protect its supply chain end, and where does retaliation for unwelcome speech begin? Lin drew that line clearly. Whether the appellate courts agree will shape how the federal government deals with every technology company that dares to say no.

Hegseth has also drawn attention for decisions that earned broad praise, such as awarding upgraded valor medals to Abbey Gate Marines. But this episode is a reminder that even well-intentioned pushback against a defiant contractor can cross constitutional lines, and that a Biden-appointed judge was more than willing to say so.

When the government starts labeling American companies as national security threats for voicing disagreements, it has stopped protecting the supply chain and started policing speech. Courts exist to catch the difference.