A federal judge in the Northern District of California ruled on August 27 that the Pentagon's designation of Anthropic as a "supply chain risk" was unlawful, permanently blocking the label in a 59-page order released August 28, according to TechCrunch, CNBC, NBC News and Nextgov/FCW. It is Anthropic's first court win in its fight with the Defense Department.

Background on the designation

Defense Secretary Pete Hegseth designated Anthropic a supply chain risk in February, marking the first time the government had publicly applied the obscure procurement-security label — typically used to flag foreign sabotage risk in military hardware and software supply chains — to a US software company. Anthropic sued in March, arguing the designation came after the company declined to let the military use Claude for domestic surveillance or in autonomous weapons systems, and that it was issued without any chance for Anthropic to contest the finding.

What the judge found

Judge Rita Lin ruled that the designation violated Anthropic's First Amendment rights by retaliating against the company for "constitutionally protected activity" — its public stance on responsible military use of AI — and separately violated its Fifth Amendment due-process rights by giving it no opportunity to respond before being labeled a risk. Lin wrote that "the empty invocation of national security is not a blank check to punish and retaliate against government critics," and pointed to the fact that the Pentagon continued pursuing other work with Anthropic even after the designation as evidence undercutting the government's stated rationale, writing that continuing to seek out the company was "not consistent with a genuine fear that Anthropic is a saboteur who would poison its software to harm national security." The ruling permanently bars enforcement of the label. The government is expected to appeal.

Why it matters

The case is among the first tests of how far the federal government can go in using national-security-adjacent procurement tools against an AI company over policy disagreements about military use of its models, rather than over a demonstrated technical or security flaw. For Anthropic, which has built a business selling Claude to federal agencies while publicly drawing lines around surveillance and weapons use, the ruling removes a stigmatizing label that could have chilled other agencies' willingness to contract with it. More broadly, it signals that courts may scrutinize similar designations closely when a company's AI safety commitments — rather than an actual security incident — appear to be the underlying trigger.