A federal judge in California has blocked the Pentagon from blacklisting Anthropic, the maker of the Claude AI models, in a ruling that strikes down the government’s designation of the company as a national-security supply-chain risk as unlawful. U.S. District Judge Rita F. Lin found that the Defense Department's measures were unconstitutional retaliation against Anthropic for publicly opposing how the military wanted to deploy its artificial intelligence technology. The decision is being read across the technology and defense industries as a landmark check on how far national-security authority can reach into the policies of an AI company, and it could reshape how the Pentagon contracts with AI developers.

Anthropic was designated a "supply-chain risk" earlier this year at the direction of Defense Secretary Pete Hegseth, after the company refused to widen military access to its Claude models for certain operations tied to mass surveillance and autonomous weapons. In a ruling delivered late Thursday evening, Judge Lin vacated the designation, concluding that the Pentagon had punished Anthropic for exercising its free-speech rights rather than for any genuine threat to U.S. security. The Trump administration is widely expected to appeal, and a second, narrower challenge from Anthropic remains pending before a federal appeals court in Washington, D.C.

Why the Pentagon designated Anthropic a supply-chain risk

Anthropic, the San Francisco-based company behind the Claude family of large language models, had become one of the most deeply integrated AI providers in U.S. defense work before the clash erupted. The Pentagon had used Claude for months as part of a roughly $200 million contract, and the model had reportedly played a role in sensitive military operations that U.S. officials publicly credited to the technology. That working relationship collapsed in the first months of 2026, when Defense Secretary Pete Hegseth pressed Anthropic to make Claude available for an unrestricted range of lawful military purposes.

Anthropic's chief executive, Dario Amodei, met with Hegseth on February 24, but the meeting did not bridge the gap. The company drew a firm line on two issues: it refused to allow Claude to be used for mass surveillance, and it declined to permit the model to be deployed in autonomous weapons that could act without meaningful human oversight. A Pentagon official responded with an ultimatum, demanding that Anthropic comply by late Friday or face consequences. When the deadline passed without compliance, the Defense Department moved to cut the company off, and Hegseth directed that contractors doing business with the U.S. military could not conduct commercial activity with Anthropic.

In March 2026, the department formalized the dispute by designating Anthropic a "supply-chain risk" under an obscure federal procurement statute originally intended to shield military systems from foreign sabotage. Legal analysts highlighted that this was the first time a U.S. company had been publicly labeled a supply-chain risk under that authority, underscoring the unusual and unprecedented nature of the move. Anthropic, for its part, said the designation would cost it billions of dollars in lost business and cause lasting reputational harm, and it warned that the fight carried implications far beyond its own balance sheet for any technology firm asked to reconcile commercial defense work with AI-safety commitments.

What the ruling found: retaliation, not security

In a detailed written order, Judge Rita F. Lin found that the Defense Department's actions against Anthropic had little to do with protecting national security and far more to do with punishing a critic. The judge concluded that the timeline of events showed the supply-chain-risk designation followed directly from Anthropic's public criticism of the Pentagon's stance on AI in the battlefield, rather than from any legitimate security finding. In the court's view, Anthropic's products did not represent a meaningful threat to U.S. national security, and the government's explanation amounted to what one legal analysis described as an "empty invocation of national security."

Judge Lin determined that the designation violated Anthropic's constitutional rights, including its First Amendment protections and its right to due process. The order characterized the government's conduct as "unlawful retaliation," and cited the administration's apparent "desire to make a public example out of Anthropic" for challenging the military's preferred approach to AI deployment. In one of the most quoted lines from the ruling, the judge wrote that the "empty invocation of national security is not a blank check to punish and retaliate against government critics," a formulation that has been widely shared across both supporters of AI safety and defense-industry analysts watching the case.

The ruling effectively blocks the government from enforcing the measures that had excluded Anthropic from U.S. defense work, clearing the way for the company to resume contracting with the military. It also carries implications beyond Anthropic alone: the decision signals that courts are willing to scrutinize whether national-security designations are being used as a tool to coerce policy concessions from technology companies. For the broader AI industry, it sets a precedent that companies may be able to set limits on how their models are used in conflict without automatically forfeiting access to the federal marketplace.

What happens next in the Anthropic-Pentagon fight

Although the ruling is a decisive legal victory for Anthropic, it is unlikely to be the final word. The administration is expected to appeal the decision, and legal observers note that appellate review could take many months and may reach the Supreme Court, given the stakes for executive authority over procurement and national security. A federal appeals court in Washington, D.C. is also still weighing a second, narrower case filed by Anthropic over the Pentagon's treatment of the company, meaning the broader dispute over military use of Claude could continue through multiple parallel proceedings.

Beyond the courtroom, the episode has prompted a broader reckoning across the defense-technology landscape. The fight put a spotlight on a tension that has defined the AI industry's relationship with the military: while Anthropic had been supplying the Pentagon for some time, it drew lines at uses it considered ethically unacceptable, including autonomous weapons without human oversight and mass surveillance. Analysts say the outcome is likely to embolden other AI companies to negotiate clearer boundaries in their defense contracts, and it has reignited public debate over how much latitude technology firms should have to refuse government work on ethical grounds.

For enterprise customers and the wider market, the practical implications are more immediate. The ruling removes a cloud that had hung over Anthropic's defense-related business and reopens the possibility of renewed military contracts for its Claude models. The decision also reinforces the message that AI governance is increasingly becoming a legal and regulatory battleground, where the boundaries of acceptable military use are being set not only by company policies but by the courts. As the appeal unfolds, companies and governments alike will be watching closely to see how far the judiciary is prepared to go in policing national-security designations, and what it ultimately means for the future of artificial intelligence in modern warfare.

Key Takeaways