A federal judge vacates the Pentagon's blacklist of Anthropic, calling it retaliation
Judge Rita Lin's 59-page order finds the "supply chain risk" designation punished Anthropic for refusing to let Claude be used in autonomous weapons and mass surveillance, not for any real security concern.
Original cover art, generated for this story. THE VISSION does not republish third-party press imagery.
- U.S. District Judge Rita Lin ruled the Pentagon's February designation of Anthropic as a "supply chain risk" was unlawful First Amendment and Fifth Amendment due-process retaliation, not a genuine security judgment.
- The dispute began when Defense Secretary Pete Hegseth objected to Anthropic's refusal to remove contractual limits on Claude's use for autonomous lethal weapons and mass surveillance of Americans.
- Lin wrote the government's actions were meant "to make a public example out of Anthropic for its 'arrogance,'" not because of any real evidence it would sabotage its model.
- The ruling vacates the designation and blocks enforcement of a related directive telling federal agencies to stop using Anthropic's technology; the government is expected to appeal, and a separate case remains pending before the D.C. Circuit.
A federal judge has vacated the Pentagon's designation of Anthropic as a "supply chain risk," ruling in a 59-page order that the label was retaliation for the company's public criticism of the Department of Defense rather than a genuine national-security judgment. U.S. District Judge Rita Lin, of the Northern District of California, found the government's conduct violated both the First Amendment and the Fifth Amendment's due-process clause.
The dispute traces to February 2026, when President Trump and Defense Secretary Pete Hegseth accused Anthropic of endangering national security and formally designated it a supply chain risk — an unusual step for a software vendor rather than a hardware supplier. Anthropic has said the real trigger was its refusal to strip two contractual restrictions from a roughly $200 million Pentagon deal: guardrails barring Claude's use in autonomous lethal weapons systems and in mass surveillance of Americans. Anthropic sued in March, calling the designation an "unlawful campaign of retaliation."
Lin's order was blunt about the government's actual motive: the designation was meant "to make a public example out of Anthropic for its 'arrogance,'" not based on any real evidence the company would sabotage its own model. The ruling both vacates the supply-chain-risk label and blocks enforcement of a related directive ordering federal agencies to cease using Anthropic's technology.
The order is a preliminary but forceful win in an ongoing fight — Anthropic's parallel case challenging related conduct is still pending before the D.C. Circuit Court of Appeals, and the administration is expected to contest Thursday's ruling. The case sits alongside a separate irony noted in coverage of the dispute: OpenAI signed its own Pentagon contract within hours of Anthropic being punished.
This is the most direct legal test yet of whether a government client can use procurement power to override an AI company's own safety guardrails, and a federal judge just said no in unusually pointed terms — calling the security rationale "empty" as a matter of law rather than deferring to executive-branch discretion, which courts often do on national-security claims. For every AI company weighing a government contract, the ruling is now evidence that refusing military end-use terms carries real legal protection, not just reputational risk.
Does the government's expected appeal overturn Judge Lin's ruling, and how does the parallel D.C. Circuit case land?
Still open. When the paper finds out, it will say so here and on the open questions page — including if it got this wrong.