The Anthropic blacklist court ruling has voided the Trump administration’s broad restrictions on the AI company, finding that officials used a national-security designation to retaliate against protected criticism. U.S. District Judge Rita F. Lin issued the 59-page decision on August 27 in the Northern District of California, largely granting Anthropic summary judgment.
Lin ruled that the government’s actions violated Anthropic’s First Amendment rights, denied the company required pre-deprivation process under the Fifth Amendment, and violated the statute used for the supply-chain-risk designation. She also found the designation arbitrary and capricious under the Administrative Procedure Act.
The court order covered actions by President Donald Trump and Defense Secretary Pete Hegseth on February 27 and March 3. They labeled Anthropic a national-security supply-chain risk, directed federal agencies to stop using its products permanently, and barred defense contractors from doing business with Anthropic, including work unrelated to the military.
What did the Anthropic blacklist court ruling actually change?
The ruling vacated those directives and, according to Ars Technica’s reporting on the order, requires the administration to rescind them. It does not compel the Pentagon, referred to in the case as the Department of War, or any other agency to buy or deploy Anthropic’s models. Lin wrote that the department remains free to choose its own AI vendor; the legal problem was the breadth and basis of the measures imposed on Anthropic.
The dispute grew out of Anthropic’s restrictions on how the government could use its technology. The company opposed uses involving fully autonomous lethal weapons and mass domestic surveillance. Administration officials, meanwhile, argued that the military needed AI access for lawful purposes, Politico reported.
Lin found the record did not establish the sabotage risk needed to support the supply-chain label. The court described the government’s administrative record as slim, with a four-page memorandum produced after two of the three challenged actions. The order says officials retreated from an earlier premise that Anthropic could gain backdoor access to its technology after it was deployed in national-security systems. Anthropic undisputedly lacked that access, the order states, and the government conceded its models were no riskier than other black-box AI systems.
That left the company’s public posture as the distinguishing factor. The government cited Anthropic’s allegedly hostile dealings through the press and its criticism of the administration’s views on AI use, according to the court order. Lin concluded that the record showed unlawful retaliation rather than a substantiated security rationale, calling the broad measures illegal and baseless.
What remains unresolved?
Anthropic filed the California case on March 9 after the restrictions were announced. Lin had temporarily blocked the measures in March; the August decision resolves most of the company’s claims on summary judgment. The court entered judgment for the government on Anthropic’s ultra vires separation-of-powers claim and certain claims involving agencies or officials that did not take relevant action or took only interim steps.
The government may appeal, The Guardian reported. Politico also reported that a separate Anthropic case in the D.C. Circuit was still pending when it published its account. Anthropic said it welcomed the ruling and intended to keep working with the government on national-security uses of AI.
This story draws on original reporting from The Verge.
