Judge rules Pentagon's Anthropic blacklist unlawful, resetting AI vendor defense rules

A federal judge has blocked the Pentagon's blacklisting of Anthropic, ruling that the government's first-ever supply-chain risk designation against a US AI company was an illegal act of retaliation. The decision strikes down a Pentagon action that threatened billions in defense contracts for the maker of Claude, and it sets a significant boundary on how the government can use procurement power to enforce military AI policy.

How the dispute led to a blacklist

The conflict began in early 2026 when Defense Secretary Pete Hegseth moved to renegotiate all AI labs' military contracts to allow the Pentagon to use AI for "any lawful use." Most AI companies signed, but Anthropic refused, drawing two red lines: no use of its models for mass surveillance of Americans and no use in lethal autonomous weapons. Anthropic CEO Dario Amodei stated that in a narrow set of cases, AI can undermine democratic values rather than defend them.

The Pentagon responded by designating Anthropic a national security supply-chain risk, a label typically reserved for foreign adversaries. The move blocked Anthropic from certain military contracts and led the Trump administration to order all federal agencies to stop using Claude. The Pentagon then signed deals with seven other AI labs including Google, Microsoft, OpenAI, and SpaceX.

The court's finding: illegal and baseless retaliation

U.S. District Judge Rita Lin issued a 59-page order finding the Pentagon's actions were "illegal and baseless." She wrote that the government's decision was based on a desire to make a public example of Anthropic for its "arrogance" in criticizing the government, not on any evidence that the company would sabotage its model. "The empty invocation of national security is not a blank check to punish and retaliate against government critics," Lin wrote.

The judge found the designation violated both the First Amendment, by retaliating against Anthropic's speech on AI safety, and the Fifth Amendment, because the company was not given due process to contest the label. The Department of War had earlier acknowledged that the designation stemmed from Anthropic's "hostile manner through the press."

Why AI builders should watch this case closely

This ruling directly tests the government's ability to use procurement labels to silence or coerce AI vendors over ethical and safety positions. For any AI lab that contracts with the US government, the message is clear: refusing to accept unrestricted military use of your models can trigger punitive procurement actions. The court found that such retaliation is unconstitutional. The practical effect is that the Pentagon must find other legal grounds if it wants to exclude a vendor based on its public policy choices, rather than using a supply-chain statute as a cudgel.

For builders shipping AI products that touch defense or intelligence workflows, the precedent matters. If the ruling holds on appeal, it strengthens the legal standing of vendors who want to set use-case restrictions without risking contract blacklisting. It also puts pressure on the DoD to articulate clearer, more substantive criteria for what constitutes a supply-chain risk from an AI company, rather than relying on vague national security assertions.

What happens next: appeals and a second case

The government is expected to appeal the ruling. Beyond this California case, Anthropic has a separate lawsuit pending in federal appeals court in Washington, D.C. over a different Pentagon supply-chain risk designation that could bar the company from civilian government contracts. That case involves a distinct legal basis and remains unresolved.

Anthropic said it remains focused on working productively with the government to harness AI for national security. But the company's stance on autonomous weapons and surveillance has not changed, and the broader legal fight over how the Pentagon treats AI vendors that speak out is far from over.