Court Rules National Security Label 'Not A Blank Check' As Anthropic Wins Landmark Retaliation Battle Against U.S. Department Of War

On August 27, 2026, Judge Rita F. Lin of the Northern District of California entered final judgment for Anthropic PBC and vacated a supply chain designation against the AI company. The court found the U.S. Department of War had unlawfully retaliated against Anthropic for publicly criticising how the government intended to use its artificial intelligence models. Judge Lin's 59-page opinion in Anthropic PBC v. U.S. Department of War, No. 26-cv-01996-RFL, converted a March preliminary injunction into a permanent one, delivering a sweeping legal victory for the AI developer. The sentence likely to define the ruling closes its introduction: 'The empty invocation of national security is not a blank check to punish and retaliate against government critics.' Judge Lin granted summary judgment to Anthropic on three separate grounds, any one of which she found sufficient to support the relief granted by the court. She held that the Presidential Directive, the Hegseth Directive, and the resulting supply chain designation constituted unlawful First Amendment retaliation, violated Fifth Amendment due process, and exceeded statutory authority under 10 U.S.C. § 3252. The factual foundation beneath all three holdings was a finding that the government's stated technical rationale collapsed entirely once the administrative record was fully assembled. The entire justification rested on a four-page memorandum from Under Secretary Emil Michael, which postdated two of the three challenged actions and relied on a premise the government later abandoned. That abandoned premise was the claim that Anthropic retained backdoor access to its deployed AI models, something the court noted Anthropic 'undisputedly lacks any such access' to. The government ultimately conceded that Anthropic's technology is 'no riskier to the national security than any other 'black box' artificial intelligence model,' leaving only an asserted loss of 'trust' as justification. Judge Lin wrote that the government's 'contemporaneous words and deeds confirm that the challenged actions were based on a desire to make a public example out of Anthropic for its 'arrogance' in criticizing the government.' The court acknowledged that 'the government is certainly owed deference on weighty issues of national security,' making its finding of improper motive all the more pointed and significant. Three further passages from the ruling carry significant weight for technology vendors operating within the federal supply chain and negotiating government contracts. On the constitutional limit, the court stated: 'Neither the Constitution nor the federal statute invoked by Defendants allows them to impose sweeping penalties based principally on Anthropic's critique of the Administration's views.' On the boundary between commercial preference and legal sanction, Judge Lin wrote: 'Though the Department of War is undisputedly free to select the AI vendor of its choice, the evidence demonstrates that the broad measures imposed on Anthropic were illegal and baseless.' The relief order vacated the § 3252 supply chain designation, struck down a directive clause barring defence supply chain companies from any dealings with Anthropic, and vacated implementing orders issued by nine agencies. Nine agencies were permanently enjoined and directed to rescind guidance and instructions they had issued, with the court retaining jurisdiction to enforce ongoing compliance with the order. Critically, the order does not compel the government to purchase Anthropic's products, and the Department remains free to decline its technology or switch to another provider through lawful means. Anthropic did not win on every argument, with the court rejecting its ultra vires separation of powers theory under Dalton v. Specter and entering judgment for agencies that took no action or only interim steps. The ruling does not resolve everything, as a parallel designation under 41 U.S.C. § 4713, the Federal Acquisition Supply Chain Security Act provision, is on direct review before the D.C. Circuit and remains in effect. That D.C. Circuit case, No. 26-1049, has been fully submitted since early August, and the San Francisco judgment does not reach or affect that separate and still-active designation. A Ninth Circuit appeal from the March preliminary injunction, No. 26-2011, has been held in abeyance since April 27, with the government's path forward now requiring a fresh notice of appeal from the final judgment itself. Anthropic filed letters under Federal Rule of Appellate Procedure 28(j) advising the D.C. Circuit of additional authorities on August 19 and again on August 28, the day after final judgment was entered. For technology vendors, the practical takeaway is clear: public advocacy on AI policy retains First Amendment protection even when it coincides with a commercial negotiating position against the government.