Judge says Pentagon turned national security into a weapon against Anthropic

Judge says Pentagon turned national security into a weapon against Anthropic
The Pentagon’s clash with Anthropic began as a dispute over how far military AI should go. It ended, for now, with a federal judge ruling that the government crossed a constitutional line.

The standoff took shape over the winter, when the Defense Department sought terms allowing its use of AI for “any lawful purpose.” Anthropic accepted military and national-security work but refused to remove two guardrails: use for mass surveillance of Americans and fully autonomous lethal weapons. Most rival labs accepted the broader terms; Anthropic’s resistance set off a public confrontation with the Trump administration.

In February, President Donald Trump and Defense Secretary Pete Hegseth branded the Claude maker a national-security and supply-chain risk, ordering federal agencies to stop using its products and barring defense contractors from doing business with it. The administration’s position was that an AI vendor should not be able to dictate how the military deploys technology it has bought, while the Pentagon also argued that opaque, enormous AI models could not be evaluated like ordinary hardware.

Anthropic sued in March, calling the action retaliation for its public safety objections. It maintained that its limits were narrow—not opposition to military operations—and that it still wanted to help the government use AI for national security.

Judge Rita Lin initially paused the blacklist. On Thursday, in a 59-page decision, she vacated the designation and found First Amendment, due-process and administrative-law violations. Her central conclusion was blunt: “The empty invocation of national security is not a blank check to punish and retaliate against government critics.”

Lin found the government’s conduct undercut its stated security fears: the Pentagon continued seeking work with Anthropic, considered treating it as essential under the Defense Production Act and collaborated on cybersecurity. The judge wrote that officials appeared bent on making an example of the company for its “arrogance,” rather than responding to evidence it could sabotage its models.

Anthropic welcomed the ruling and said it remained focused on working productively with Washington. The administration is expected to appeal, and a separate challenge to another Pentagon designation remains pending in the D.C. Circuit.

Continue reading https://foxvector.com/stories/01a04b30-00ee-2e2d-7063-3fa7734558dc