Summary
- Judge Rita Lin issued a 59-page ruling granting Anthropic summary judgment on claims related to the First Amendment, due process, and the Administrative Procedure Act.
- The government acknowledged that Anthropic does not have backdoor access to its models and poses no greater risk than any other AI system.
- However, Anthropic's argument that Trump's order exceeded presidential authority was dismissed.
A federal court has ruled that the Trump administration acted unlawfully in retaliating against Anthropic after the company publicly declined to permit military applications of its Claude language model for purposes like mass surveillance and autonomous weaponry.
The conflict arose when the Pentagon requested that Anthropic remove all restrictions on usage and accept a clause allowing "all lawful uses." While Anthropic complied with most requests, it maintained two critical restrictions. On February 27, Trump ordered federal agencies to cease using Anthropic's technology, labeling the company a "RADICAL LEFT, WOKE COMPANY" on Truth Social. Following this, Hegseth accused Anthropic via Twitter of "a master class in arrogance and betrayal" and prohibited any military contractor from engaging with the company.