A US federal judge on Thursday ruled that the Trump administration acted unlawfully by designating AI company Anthropic a national security supply-chain risk, overturning measures that had restricted the company’s access to US government and military contracts.

US District Judge Rita Lin found the Pentagon’s designation of Anthropic “illegal and baseless” in a 59-page order, ruling that the government had unlawfully retaliated against the company over its position on military uses of artificial intelligence. The decision makes permanent an earlier temporary suspension of the measures and takes effect immediately, although the government can appeal.

Lin said national security could not be used to penalise companies for challenging government policy. “The empty invocation of national security is not a blank check to punish and retaliate against government critics,” she wrote.

The dispute began in February after Anthropic refused to allow its Claude AI models to be used for US surveillance or autonomous weapons, citing concerns about the reliability and safety of the technology. Defense Secretary Pete Hegseth subsequently designated Anthropic a supply-chain risk, a status intended to protect military systems from potential foreign sabotage.

The designation was the first public use of the supply-chain risk classification against a US company, according to Reuters, and prevented Anthropic from securing certain military contracts. Anthropic executives have said the measures could cost the company billions of dollars in lost business and reputational damage.

Anthropic challenged the designation in March, arguing that the administration had violated its First Amendment right to free speech by retaliating against its views on AI safety and had denied it due process under the Fifth Amendment. The company said the decision was unsupported by facts and inconsistent with the military’s previous use and praise of Claude.

The Justice Department argued that the restrictions arose from Anthropic’s refusal to accept contractual terms rather than its views on AI safety, according to court filings. Government lawyers said the uncertainty over how the Pentagon could use Claude could create operational risks, including the possibility of military systems being disabled during operations.

Anthropic welcomed Lin’s decision, saying it would continue working with the government. An Anthropic spokesperson told the BBC that the company remained “focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology”.

The ruling does not resolve a separate case brought by Anthropic in Washington DC concerning another Pentagon supply-chain risk designation that could affect its ability to secure civilian government contracts. The White House and Pentagon had not immediately commented on the California ruling.

---