A federal judge has ruled in favor of Anthropic, saying the Pentagon’s actions against the AI company were unlawful and motivated by a desire to punish public criticism rather than by any credible security concern.
Judge Rita Lin issued a written order that the department’s measures were “illegal and baseless,” a finding that centres on the February decision by the administration to designate the company a supply chain risk after public disputes over whether Anthropic’s models should be used for fully autonomous weapons or mass surveillance. The government is expected to fight the ruling.
The clash escalated when President Donald Trump and Defence Secretary Pete Hegseth publicly accused Anthropic of endangering national security and Anthropic CEO Dario Amodei refused to back down from the company’s red lines. In the judge’s account, the department’s penalties “were based on a desire to make a public example out of Anthropic for its ‘arrogance’ in criticising the government, not based on any articulable basis to believe that Anthropic would actually sabotage its model,” Lin wrote in the 59-page ruling.
Anthropic sued the Pentagon in March, calling the actions an unlawful campaign of retaliation; the court had earlier temporarily blocked the Pentagon from enforcing the supply-chain label and from implementing a social-media directive that barred federal agencies from using Anthropic’s chatbot. Anthropic also has a separate, narrower case pending in the federal appeals court in Washington, D.C., that challenges a different rule the department used to justify the designation sued the Pentagon in March.
The written decision followed a July hearing in which Lin described the government’s stance as “really troubling” and at odds with the First Amendment. At that hearing, Justice Department lawyers argued that large AI models are unusually opaque and therefore hard to evaluate, a technicality the court found insufficient to justify the sweeping penalties. Anthropic’s lawyer Michael Mongan had warned the measures would “profoundly harm Anthropic” and chill debate on the issue In a hearing July 30.
The dispute took on practical and commercial stakes: the ruling notes that rival firms moved quickly after Anthropic’s public dispute with the Pentagon, with OpenAI reaching its own Pentagon agreement hours after the government punished Anthropic, and both companies preparing for high-profile public offerings ramping up for buzzy initial public offerings.
An Anthropic spokesperson said the company welcomed the judgement and remained focused on working with government partners to harness AI for national security. The judge’s finding narrows the Pentagon’s room to impose punitive supply‑chain labels where the record shows retaliation rather than risk; the next step is a likely government appeal.
