Judges on the U.S. Court of Appeals for the District of Columbia Circuit on Tuesday appeared divided during oral arguments in a legal dispute between the Pentagon and artificial intelligence company Anthropic, which claims Defense Secretary Pete Hegseth unlawfully branded it a national security risk for raising ethical concerns about AI use in war.
The three-judge panel — Judges Karen LeCraft Henderson, Sharon Swingle, and Gregory Katsas — heard Anthropic’s appeal of a lower court decision that allowed the Defense Department’s “supply-chain risk” designation to stand. Henderson, an appointee of Republican President George H.W. Bush, said during arguments that she sees no evidence to support the Pentagon’s determination.
“To me, this is just a spectacular overreach by the [Defense] Department,” Henderson said.
Anthropic, a San Francisco-based AI safety company, publicly raised concerns about the military’s use of artificial intelligence in combat, including the potential for autonomous weapons and algorithmic decision-making in lethal operations. After those statements, the Pentagon designated Anthropic a supply-chain risk under federal acquisition regulations, a classification that can restrict a company’s ability to contract with the Defense Department.
The company sued Hegseth in his official capacity, arguing that the designation was retaliation for its protected speech and lacked factual basis. The Pentagon has defended the label, maintaining that Anthropic’s public posture on military AI could affect its reliability as a contractor.
Judges Swingle and Katsas did not reveal their leanings during the hearing, but their questions suggested the court is wrestling with how much deference to give the Pentagon’s security assessment and whether the company’s speech is protected commercial advocacy or a legitimate factor in procurement decisions.
The panel did not indicate when it would issue a ruling.