The ruling keeps a separate legal challenge alive as CEO Dario Amodei defends limits on military uses of Claude and Anthropic contests the designation.
By [RuntimeWire Staff](https://runtimewire.com/author/runtimewire-staff)
· Published
Primary source: [Reuters](https://www.reuters.com/world/us-appeals-court-declines-block-pentagons-blacklisting-anthropic-2026-09-25/)
Why it matters #
The dispute turns Anthropic's safety commitments into a commercial constraint: the safeguards that define its position can also limit access to defense contracts, while separate court cases produce different near-term outcomes.
A federal appeals court on Friday, September 25th, declined to block the Pentagon's designation of Anthropic as a national-security supply-chain risk, leaving the challenge by Anthropic, led by CEO Dario Amodei, alive while a separate California case has blocked the designation. The Reuters report says Anthropic argues the action has cost it billions in lost business and damaged its reputation ahead of a potential initial public offering.
For Dario Amodei (@DarioAmodei), Anthropic's co-founder and CEO, the dispute tests a position he has stated publicly: the government should use AI for national security, while Anthropic should retain limits on certain uses of its models. In a February statement, Amodei said Anthropic was willing to support the Department of War and intelligence agencies, but would not remove its safeguards against mass domestic surveillance and fully autonomous weapons. He argued that current AI systems are not reliable enough to select and engage targets without human involvement. Those limits, rather than a refusal to work with the military, sit at the center of the conflict.
Two courts, two legal tracks
The D.C. Circuit's decision concerns a case Anthropic brought directly against the Pentagon's supply-chain-risk designation. The designation followed a dispute over whether the department could require Anthropic to permit uses of Claude that the company had excluded from its contracts. In its February statement, Amodei said Anthropic had offered to work with the department on research to improve autonomous systems, while maintaining its position that the safeguards should remain.
A separate case in California has produced a different result. On August 27th, a federal judge granted Anthropic summary judgment on several claims challenging the government's actions, including its First Amendment and due-process claims and its challenge to the supply-chain designation. The court's order describes that case as covering a broader set of government actions. Reuters reported that the designation had been blocked in that proceeding.
The two outcomes leave Anthropic with meaningful relief in California and a setback in its separate D.C. Circuit case. They should not be read as a single court issuing opposite final rulings on the same question: the cases proceed through different legal routes and address government actions under distinct claims. The September 25th decision, as Reuters describes it, declines to block the Pentagon's action in the D.C. case; it does not by itself resolve the underlying dispute over whether the designation was lawful.
Amodei's boundary is also a business decision
Amodei's case is more complicated than a simple rejection of military work. Anthropic's statement said Claude was already deployed for national-security work, including intelligence analysis, operational planning and cyber operations. Anthropic said it wanted to continue serving the department while retaining two restrictions. The Pentagon, in turn, has demanded that AI contractors accept use for any lawful purpose, according to Amodei's statement.
That line affects how Anthropic can sell Claude to the government and to businesses that perform defense work. In an earlier account of the case, the D.C. Circuit's April order said the Pentagon had canceled its Anthropic contracts and barred its other contractors from using Anthropic as a subcontractor on work performed for the department. The order also said the restriction did not bar those contractors from using Claude on work for other customers. The practical reach of the designation is therefore tied to defense work, even as Anthropic says the consequences extend to lost business and reputation.
That claimed loss deserves careful handling. Reuters attributes the billions-of-dollars figure to Anthropic; it is not an independently established loss estimate in the report. Anthropic has not put a precise figure on the alleged damage in the supplied reporting. Reuters has separately reported that Anthropic confidentially filed for an IPO in June, giving the dispute added weight as Anthropic prepares for public-market scrutiny. The designation's effect on revenue, customer relationships and the IPO case remains distinct from the legal question of whether the government acted within its authority.
The case also puts pressure on Anthropic's founding identity. Anthropic describes itself as an AI research and product company focused on making systems reliable, interpretable and steerable; its leadership page identifies Amodei as a co-founder and CEO who leads its research direction. That safety thesis has commercial value when customers want a model provider willing to set and explain boundaries. In defense procurement, those same boundaries can become a point of direct conflict with the buyer.
The next consequential issue is whether the two legal tracks converge on how much control a model developer can retain over military use after delivery. The September 25th ruling gives Anthropic no relief in the D.C. Circuit at this stage. California's decision gives it a substantial win in a separate challenge. For Amodei, the unresolved balance is concrete: preserving the conditions under which Anthropic will supply Claude, while defending the company's ability to compete for government and defense-adjacent business.