Background
Anthropic filed its petition for review on 9 March 2026, challenging the Department of War’s determination under 41 U.S.C. § 4713 that Anthropic presents a supply chain risk to national security. The petition states that the Secretary publicly directed the designation on 27 February 2026 and that, on 4 March 2026, Anthropic received a notice dated 3 March 2026 stating that the use of Anthropic’s products or services in covered Department procurements presents a supply chain risk, that less intrusive measures were not reasonably available, and that the action was effective immediately across all covered products and services. Anthropic’s emergency motion then placed that action in context, explaining that the company had been supplying Claude for defence and classified use, but that the relationship broke down after Anthropic refused to remove two remaining restrictions, namely use for lethal autonomous warfare and mass surveillance of Americans, while agreeing to other national-security uses. Anthropic now characterises the designation as retaliatory, procedurally unlawful, arbitrary, and beyond statutory power (1208828684, 1208829653).
AI Interaction
The AI issue is central and direct. This case is about whether a frontier AI developer may retain deployment guardrails when supplying models for military and national-security use, and whether the government may treat those retained limits as evidence that the vendor itself has become a supply chain risk. Anthropic presents Claude as a powerful model used in sensitive settings but still subject to narrowly maintained usage restrictions grounded in safety concerns and responsible deployment. The government presents the matter differently, arguing that AI systems are unusual because they require continuing vendor support, tuning, and trust, and that Anthropic’s conduct created the risk that it could manipulate, withhold, or constrain software embedded in highly sensitive defence systems according to its own policy judgments. The dispute therefore raises a direct conflict over who controls the operational boundaries of deployed AI in national-security settings. (1208829653, 1208832389, 1208833581).
Notes
- Anthropic’s petition itself states that a related challenge concerning 41 U.S.C. § 3252 is being litigated separately in the District Court, N.D. California, in Anthropic PBC v. U.S. Department of War, No. 3:26-cv-01996, while this D.C. Circuit case directly challenges the § 4713 designation (1208828684).
- The immediate next development to watch is the D.C. Circuit’s ruling on the emergency stay motion (1208830145 1208833581).