Pentagon Anthropic Supply Chain Risk Ruling: What Changes
Published by AINave Editorial • Reviewed by Ramit
The Pentagon Anthropic supply chain risk ruling leaves the Defense Department’s restrictions on Claude in place, but it does not settle every legal challenge to the government’s actions against Anthropic. On September 25, 2026, a D.C. Circuit panel ruled 2-1 that the Pentagon had enough support to designate the company a supply-chain risk.
The designation bars the U.S. military from using Anthropic’s models and blocks defense contractors from using them in work with the Defense Department. The panel delayed its decision from taking immediate effect, giving Anthropic time to seek rehearing or pursue further review. Anthropic said it was considering its options, including further review.
The dispute was about control over model use
The ruling followed negotiations over Claude’s potential deployment on the Pentagon’s GenAI.mil platform. The department wanted access for all lawful purposes; Anthropic sought assurances that Claude would not be used for fully autonomous weapons or domestic mass surveillance. The talks failed, after Anthropic had signed a $200 million Pentagon contract in July 2025.
The majority’s concern was not simply that Claude had safety limits. It said the government had ample support for viewing the model’s continued integration into Defense Department systems, including through contractors, as a national-security risk. The court pointed to restrictions Anthropic encoded into Claude and the possibility, in the government’s view, that they could prevent tasks the department considered authorized and necessary. That is the court’s account of the Pentagon’s rationale, not a general technical finding that model safeguards create security risks.
This makes the case consequential for procurement: a supplier’s built-in limits on use became part of the government’s risk argument. The ruling does not establish that every restriction on an AI system will justify a similar designation; it upholds this designation on the record before the court.
One designation upheld, another challenged
The D.C. Circuit case addressed one of two distinct designations the Pentagon relied on. A federal judge in San Francisco had found the other designation unlawful, while the D.C. Circuit upheld the one before it. The two outcomes therefore coexist rather than cancel each other out. The appeals panel allowed time for Anthropic to seek a rehearing, including before the full D.C. Circuit, or ask the Supreme Court to take the case.
For teams working on Defense Department projects, the immediate practical boundary is specific: the reported restriction covers military use and contractor use in work with the agency. The legal position could still change through further proceedings, but the ruling underscores how model-level controls can become a procurement issue when a customer expects broad discretion over lawful use.
