In a significant legal setback for the Pentagon, a federal judge in California has temporarily halted the government's campaign against AI company Anthropic. Judge Rita Lin issued a preliminary injunction on Thursday, siding with Anthropic's argument that the Defense Department likely overstepped its authority.

The dispute centers on Anthropic's refusal last year to let the Pentagon use its Claude AI for autonomous weapons or domestic surveillance. In response, the Defense Department labeled Anthropic a "supply chain risk," triggering a directive for agencies to stop using its technology. Anthropic sued, claiming this violated its First Amendment rights by punishing it for setting ethical boundaries.

Judge Lin's ruling suggests the government's actions appear "arbitrary and capricious." During a hearing, she questioned why the Pentagon would issue the severe designation instead of simply ending its contract. "It looks like an attempt to cripple Anthropic," Lin stated from the bench. She noted the government provided no evidence that Anthropic's stance made it a security threat.

The injunction freezes the punitive measures for at least a week as the case proceeds. It also complicates a massive, costly effort to rip Claude out of government systems, where it's reportedly used for tasks like target analysis. The ruling casts doubt on the legal foundation of the administration's pressure campaign against a leading AI firm for its public policy positions.

Source: The Guardian