Court Rules Trump Administration Illegally Blacklisted Anthropic

Introduction

A federal judge has ruled against the Trump administration in a months-long dispute with Anthropic, finding that the Pentagon’s decision to label the AI company a “supply chain risk” was unconstitutional. Rita F. Lin, a district judge in the Northern District of California, said the measure was unsupported and amounted to unlawful retaliation under the First Amendment.

The dispute was not simply about whether Anthropic could win or keep a defense contract. It raised a broader question: can the government use a national-security designation to pressure an AI company after it publicly maintains limits on how its systems may be used by the military?

Key points

• The Pentagon sought to renegotiate existing contracts with AI companies so that its systems could be used for “any lawful use.”
• Most labs accepted the revised terms, while Anthropic retained two restrictions: no mass surveillance of Americans and no lethal autonomous weapons operating without human oversight.
• After Anthropic refused to change its position, pressure from administration and Defense Department officials escalated. The company was then designated a supply chain risk, a classification generally associated with national-security threats.
• The Pentagon moved to reduce Anthropic’s influence by signing agreements with seven other AI companies and labs, including Google, Microsoft, OpenAI, and SpaceX.
• Anthropic filed suit in March. Judge Lin had already temporarily blocked the designation; the latest ruling confirms that the underlying action was unlawful.

Why the court rejected the designation

Judge Lin acknowledged that the Defense Department is free to choose the vendors it wants to work with. That discretion, however, does not allow officials to punish a company for criticizing the government’s contracting position. The court highlighted records indicating that Anthropic’s designation was linked in part to what officials described as the company’s “hostile manner through the press.”

In the judge’s view, using that public stance as a reason for punitive action was a classic example of First Amendment retaliation. The court also rejected the idea that invoking national security automatically settles the constitutional question. The government may assess technology and supply-chain risks, but it still needs evidence and a lawful decision-making process. Broad security language cannot serve as a blank check for measures aimed at a particular speaker.

What the ruling means

The decision offers an important marker for the relationship between public procurement, AI safety rules, and corporate speech. AI companies may continue to establish boundaries around surveillance, weapons, and human oversight without automatically surrendering the possibility of government cooperation. At the same time, agencies that respond to those boundaries with special sanctions could face judicial scrutiny.

For the Pentagon, the ruling makes future efforts to authorize AI for “any lawful use” more demanding. Officials will need to explain the operational need for such terms, show that their evidence supports the action, and demonstrate that the response is not punishment for a company’s stated position.

Anthropic said it welcomed the ruling and remained focused on working productively with the government on national-security applications. The case does not settle every question about military AI, but it establishes a significant principle: policy disagreements over high-risk systems should be addressed through negotiation and transparent rules, not converted into a supply-chain ban because a company has publicly drawn safety lines.

Source: The Verge AI