Anthropic’s Pentagon blacklist was ruled unlawful on Thursday, 27 August 2026, when a California federal judge set aside the designation and permanently barred the United States government from enforcing it.
US District Judge Rita Lin, sitting in the Northern District of California, found that the Pentagon had retaliated against the AI company and denied it the process it was owed, as reported by AP News.
A permanent injunction followed, the strongest remedy a court could issue short of damages.
What the Anthropic Pentagon blacklist actually did
A supply chain risk designation was the federal government’s way of labelling a supplier too dangerous to buy from.
Once applied, it let agencies strip a company out of defence contracts and procurement pipelines without the usual bidding process, effectively cutting it off from the largest technology customer on earth.
Defence Secretary Pete Hegseth applied that label after Anthropic refused to let the military point Claude, its chatbot, at United States surveillance work or autonomous weapons.
The company’s published usage policy had long ruled out both, and Anthropic declined to carve out an exception for a customer with a very large budget.
Inside Judge Lin’s Anthropic Pentagon blacklist ruling
In her judgment, Lin set out two separate constitutional failures, the First Amendment protection against being punished for what you say, and the Fifth Amendment protection that requires a hearing before the state takes something away from a party.
Judge Rita Lin wrote:
“The undisputed record shows that the challenged actions constituted unlawful retaliation in violation of the First Amendment, and that Anthropic was denied the pre-deprivation process required under the Fifth Amendment.”
The second finding went at the designation itself rather than the motive behind it. Arbitrary and capricious was the standard American courts used when an agency reached a decision with no rational basis in the evidence or in the statute it was meant to follow.
Lin also found:
“Secretary Hegseth’s decision to designate Anthropic a supply chain risk violated the governing statutory scheme…and was arbitrary and capricious.”
Where the ruling leaves Anthropic and the Pentagon
With the designation set aside, Anthropic returned to the position it held before the label was applied, free to bid for federal work without a formal risk marker hanging over it.
The permanent injunction meant the government could not simply reissue the same designation under a new file number.
The clash was not Anthropic’s first collision with Washington. In June 2026 the company spent 19 days with its most capable models switched off worldwide after the US imposed export controls on Fable 5 and Mythos 5, a sign of how tightly frontier AI had become entangled with national security policy.
The government still had the option of an appeal to the Ninth Circuit, the federal court that covers California, which could put the injunction back in play. In the meantime Hegseth’s department was left buying its AI from a supplier that publicly refused to sell it surveillance and weapons capability.
