A U.S. district judge has blocked the Pentagon from treating Anthropic as a national security supply-chain risk. The order stops enforcement of a blacklist placed on the maker of Claude by Defense Secretary Pete Hegseth.
Background on the designation
Anthropic filed suit in California federal court after the department applied the supply-chain risk label. That label allows the government to restrict a company’s access to military contracts and systems on grounds of potential infiltration or sabotage by adversaries. The company argued the move was aimed at its positions on AI safety in battlefield use rather than any documented supply-chain defect. Prior to the ruling, the designation stood and limited Anthropic’s dealings with Defense Department programs.
The dispute traces to broader tensions between the Pentagon and commercial AI developers over how frontier models may be used in combat settings. Anthropic has publicly advocated limits on autonomous weapons and other lethal applications. The department responded by invoking supply-chain authorities that normally address hardware or software components known to carry foreign adversary links.
The court’s order
U.S. District Judge Rita Lin, appointed by former President Joe Biden, issued a 59-page order on Thursday. She concluded the Pentagon’s decision was “illegal and baseless.” The ruling states: “The empty invocation of national security is not a blank check to punish and.” Anthropic’s complaint centered on the claim that Hegseth overstepped authority when he applied the supply-chain risk tag. The suit described the action as an attempt to pressure the company over its public stance on limiting autonomous weapons and other AI applications in combat. No evidence of actual infiltration or sabotage risk was presented in the department’s filing, according to the court record.
The order prevents the Pentagon from using the blacklist designation while the case proceeds. Anthropic had sought the injunction to preserve its ability to work with military customers on non-combat AI tools. The department had cited broad national security concerns tied to the company’s technology reaching potential adversaries through supply chains, but the judge found those concerns lacked concrete grounding in the administrative record.
Judge Lin’s opinion examined the statutory basis for the designation and found it did not meet the requirements set by Congress for supply-chain risk actions. The ruling distinguishes between legitimate security reviews that rest on specific intelligence and broader policy disagreements that cannot be reframed as supply-chain threats without additional factual support.
Immediate effects
Anthropic can now maintain existing engagements and pursue new ones with Defense Department components that do not involve weapons systems. The preliminary injunction applies only during litigation and does not resolve the underlying lawsuit. Both sides retain the ability to present further evidence as the case advances through discovery and potential trial.
The decision arrives at a moment when multiple AI labs are negotiating access terms with military programs. A blocked designation removes one immediate barrier for Anthropic while leaving open the possibility that the Pentagon could pursue other regulatory or contractual avenues.
Why it matters
The ruling limits the Defense Department’s flexibility to sideline AI vendors when the core disagreement concerns acceptable use policies rather than verifiable technical vulnerabilities. Companies building large language models now have a documented precedent for challenging supply-chain risk labels in court when the administrative record contains no concrete evidence of adversary access. Military teams seeking commercial models for logistics, intelligence analysis, or training applications gain a narrower window in which policy disputes can cut off an entire vendor. The opinion also signals that judges will require the department to tie any future designation to specific facts rather than general assertions of national security risk. Anthropic remains able to pursue non-combat work with defense customers while the lawsuit continues, preserving at least one frontier model provider in that segment of the market.
---
Sources:
{"word_count": 682, "sources_used": 1, "expanded_sections": ["context", "detail", "why_it_matters"]}
No comments yet