Federal Judge Strikes Down Pentagon's 'Supply Chain Risk' Label on Anthropic
A California federal judge ruled that the Trump administration's designation of Anthropic as a national security supply chain risk was unlawful, citing First Amendment retaliation and lack of due process. The court found the label arbitrary and capricious, noting contradictions such as the Pentagon continuing to pursue contracts with the AI firm. The ruling marks Anthropic's first legal victory in its ongoing disputes with the Department of Defense over AI safety guardrails.
Judge Rita Lin's ruling highlighted internal contradictions in the government's position, noting that Defense Secretary Pete Hegseth simultaneously proposed treating Anthropic as essential to national security under the Defense Production Act while labeling the company a supply chain threat. The court also observed that the Department of Defense continued pursuing contracts with Anthropic and collaborated with the firm's Mythos model for cybersecurity work, undermining the basis for the ban.
Anthropic's legal challenge began in March with two separate complaints filed in California and Washington, D.C. The California case has now concluded in the company's favor, while the D.C. proceedings remain active. The underlying conflict traces to Anthropic's refusal to permit its models to be used for fully autonomous weapons or mass surveillance of American citizens, positions the company maintained despite Pentagon assurances about lawful use.
This ruling could reshape how AI companies negotiate with federal agencies, potentially emboldening firms to set safety conditions without fear of government retaliation. If courts continue to scrutinize national security designations, agencies may face greater accountability in procurement decisions. The decision may also influence public trust in AI governance, signaling that companies can challenge executive actions through judicial review. However, the ongoing D.C. case and potential appeals could limit the ruling's immediate practical effect on defense-AI relationships.