Anthropic asks full appeals court to rehear its Pentagon blacklist case
A split panel let the Pentagon bar Claude from military systems. Now Anthropic wants all the judges on the D.C. Circuit to look again, while a separate federal judge has already called the same Pentagon campaign illegal.
Anthropic filed a petition on October 13 asking the full U.S. Court of Appeals for the D.C. Circuit to rehear its challenge to the Department of War's decision to designate the company a national security supply-chain risk, according to Bloomberg Law. The filing follows a September 25 ruling: a three-judge panel voted 2-1 to uphold the designation, which bars the Pentagon and its contractors from using Anthropic's Claude models in work for the department.
The fight traces back to February 27, when President Trump ordered federal agencies to immediately cease using Anthropic's technology. The Pentagon allowed a transition period of up to 180 days to remove Claude from department systems. Defense Secretary Pete Hegseth made the formal determination on March 3. Anthropic received notice the next day: the department was designating Anthropic and all its products a supply-chain risk under the Federal Acquisition Supply Chain Security Act of 2018, a law shaped by earlier supply-chain security controversies involving companies such as Kaspersky. The timing was not an accident. Anthropic CEO Dario Amodei had publicly refused to loosen the guardrails in Claude's usage terms that block the model from being used in autonomous weapons systems or domestic surveillance, and the Pentagon wanted those restrictions gone.
Judge Gregory Katsas wrote the panel's 43-page majority opinion, joined by Judge Neomi Rao, finding that the Department of War acted within its statutory authority under FASCSA when it issued the designation without a pre-decision hearing. Judge Karen LeCraft Henderson dissented in an eight-page opinion, arguing the designation came "without any pre-designation process and despite straightforward evidence the designation was retaliation for Anthropic's speech about AI safety," as Anthropic's own petition for rehearing now repeats almost verbatim.
Here's the part that makes this case genuinely strange: a different federal judge already ruled the opposite way on nearly the same facts. In a separate suit Anthropic filed in California, U.S. District Judge Rita Lin found on August 27 that "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." Lin went further, calling Hegseth's decision arbitrary and capricious under the governing statute. Two federal courts, looking at the same designation letters and the same refusal to drop safety limits, reached opposite conclusions about whether the Constitution was even implicated.
The Trump administration designated Anthropic a national security supply-chain risk, forced it to recall its Mythos AI model under export controls, and banned federal agencies from using its products. While OpenAI and Google secured federal contracts by playing the Washington access game, Anthropic's refusal to allow autonomous weapons use has... - why did Trump ban Anthropic - Anthropic supply chain risk designation Pentagon
The California ruling is a win on paper. But FASCSA designates the D.C. Circuit as the exclusive venue for challenging a supply-chain risk order. That's exactly why Anthropic's rehearing petition there carries more practical weight than Lin's injunction. If the full D.C. Circuit declines to rehear the case, or rehears it and sides with the Pentagon again, the designation stands regardless of what the California court found. En banc review is discretionary, and the D.C. Circuit grants it rarely. Anthropic's path after that runs through a cert petition at the Supreme Court, which is equally discretionary and far from guaranteed.
What the designation actually does in practice is blunt. FASCSA clauses bar the use of Claude in covered contract performance. They require contractors to monitor for it, and force reporting within three business days if Claude shows up anyway, according to law firm analyses of the order from Mayer Brown and Holland & Knight. Anthropic has spent two years courting government and defense customers, including a Pentagon contract of its own announced in July 2025. Being formally walled off from that market by the department it was trying to sell into is a direct hit to a real revenue line, not an abstract reputational problem.
It also sits awkwardly next to Anthropic's public identity. The company has built its brand on being the safety-conscious lab, the one that turns down business it considers dangerous. That stance is now the specific thing the government cited as the reason for the ban. OpenAI, meanwhile, struck its own Pentagon deal within hours of Anthropic's punishment becoming public, a contrast Anthropic's lawyers have not been shy about pointing out in court filings.
Trump has rebranded the Pentagon as the Department of War, authorized as a secondary title by executive order in September 2025 and reflected in the war.gov domain now used alongside Defense Department channels. Same department, old FASCSA authority. The name change is cosmetic. The underlying agency power to bar an AI vendor from military systems is not, and Anthropic's October 13 filing is the clearest sign yet that the company thinks the September ruling got the legal standard wrong, not just the outcome.
XBOW researchers used AI agents to win a $250,000 Chrome exploit bounty • Liquid AI's LFM2.5-24B-A2B Runs Agent Tasks Locally on a Laptop • Figure AI's former safety engineer says its home robot can fracture a skull
This article is posted in AI News, check it out for more related stories.
When the Trump administration banned Anthropic from federal use in February 2026 and OpenAI announced a Pentagon deal hours later, the sequence exposed something direct: Washington now picks AI vendors through political relationships as much as technical merit. With Cohere merging with Aleph Alpha at a $20 billion valuation and the Pentagon's... - how Washington picks AI winners - federal government AI policy preferences
