Mastodawn

undefined | Anthropic Supply-Chain Risk Label Should Stay In Place, Appeals Court Says by Paresh Dave

Anthropic Inc. lost a bid to have the Pentagon’s “supply‑chain‑risk” label temporarily removed when a three‑judge U.S. Court of Appeals in Washington, DC, ruled on Wednesday that the AI firm had not satisfied the strict requirements for relief. The appellate decision directly contradicts a San Francisco district‑court ruling issued a month earlier, which found the Department of Defense had acted in bad faith and ordered the label erased, prompting the Trump administration to restore Anthropic’s access to its Claude models across the federal government. The appellate panel stressed that, even if the designation harms Anthropic financially, overturning it could force the military to continue working with a vendor it deems unsafe during an ongoing conflict.

The two courts are each addressing a different legal mechanism the Pentagon used to impose the same practical restriction, making Anthropic the first U.S. company flagged under both supply‑chain‑risk statutes that are normally reserved for foreign entities deemed national‑security threats. Anthropic argues that the label unfairly penalizes it for refusing to let its Claude system be used in high‑risk operations—such as fully autonomous drone strikes—without human oversight, and it claims the government’s action has cost it lucrative contracts. The company’s spokesperson, Danielle Cohen, said the Washington decision shows the urgency of the issue and expressed confidence that the courts will eventually deem the designations unlawful, while the DoD has not commented.

Legal scholars note that the case tests the limits of executive power over technology firms and could set a precedent for how the government regulates AI in defense. Experts familiar with government contracting say Anthropic has a strong argument, yet courts often defer to the White House on national‑security matters, a deference that could chill broader debate about AI performance and safety. Final rulings on the two lawsuits may take months, with oral arguments in the D.C. court scheduled for May 19, leaving the future of Anthropic’s role in Pentagon AI projects—and the broader landscape of AI governance—still uncertain.