Judge pauses Pentagon move to block Anthropic AI tools

In short: Anthropic says the US government ordered agencies to stop using two of its Claude AI systems for national security reasons, but a federal judge has paused the Pentagon’s blacklist while a lawsuit continues.

What happened

Anthropic, a San Francisco AI company known for its Claude models, says the US federal government ordered a suspension of two Claude based AI systems used in government work. The company said the order cited national security concerns.

Court filings and reporting say the Department of Defense labeled Anthropic a “supply chain risk to national security.” In simple terms, that label can act like a no entry sign, making it hard or impossible for a company to sell to the Pentagon and often to other parts of the federal government.

Anthropic argues the label is retaliation. The company says it would not agree to a Pentagon contract unless the deal included promises that its AI would not be used for autonomous weapons (weapons that can select and attack targets on their own) or for mass surveillance of Americans.

Anthropic sued the government and asked a court for a preliminary injunction, which is a temporary order that pauses a government action while the case is decided. A federal judge in San Francisco has now blocked or indefinitely paused the Pentagon’s attempt to treat Anthropic as a supply chain risk, saying officials likely broke the law and retaliated over the company’s public stance.

Why it matters

This fight is about who controls the rules for AI used by the military and intelligence agencies. It could affect whether AI companies can set binding limits on how their tools are used, even when big government contracts are involved.

Source: NYTimes