A federal judge says the Pentagon can't blacklist Anthropic for refusing to build weapons on demand.
Anthropic sued the Trump administration in March, arguing the Pentagon had retaliated against the company for setting 'red lines,' restrictions on how the Pentagon could use its AI models. On Thursday, US District Judge Rita F. Lin ruled the blacklisting unconstitutional, closing out a months-long legal fight in Anthropic's favor. Lin wrote that invoking national security doesn't give the government a free pass to punish critics. The ruling doesn't detail which specific contracts or programs triggered the blacklist.
This is a rare case of an AI company beating the government in court over how its technology gets used, instead of settling the dispute quietly. It also tests how far the administration can lean on national security as justification when a contractor pushes back on military use cases, a tension likely to resurface as more AI labs set their own restrictions. For Anthropic, which has been more vocal than rivals about limiting military applications, the ruling validates a stance that carried real business risk.
Lin's ruling put it plainly: 'The empty invocation of national security is not a blank check to punish and retaliate against government critics.'