Court Overturns Pentagon's AI Vendor Ban as Unlawful Retaliation
A federal judge in California vacated the Defense Department's designation of Anthropic as a national security supply-chain risk, calling the action illegal and baseless. The ruling also lifted related sanctions imposed by nine agencies, though it does not force the Pentagon to use Anthropic's models. The dispute originated from a $200 million contract involving Claude AI and a military operation in Venezuela.
The ruling stems from a contract dispute that began after reports emerged of Claude's use in the operation that captured Venezuelan president Nicolás Maduro. Anthropic sought restrictions on deployments involving lethal autonomous weapons and mass surveillance, but Hegseth rejected those conditions, insisting the contract permitted "all lawful use." When negotiations collapsed in February, the supply-chain risk designation followed.
Anthropic pursued two legal challenges simultaneously, one in California federal court and another at the DC Circuit. The California ruling vacated the designation and lifted sanctions from nine agencies, though the Pentagon retains discretion to choose other AI vendors. The DC case remains pending, and the Defense Department is expected to appeal Thursday's decision.
This ruling could reshape how AI companies engage with military contracts, potentially emboldening vendors to negotiate usage restrictions without fear of retaliation. It may also influence how the Pentagon approaches AI procurement, possibly making it more cautious about punitive designations. Other AI firms watching this case could gain confidence in pushing back on government demands, though the expected appeal leaves the ultimate legal outcome uncertain. The decision may affect national security technology partnerships broadly.