Why a Judge Blocked the Trump Administration’s Anthropic Ban

A federal judge has temporarily blocked the Trump administration from banning Anthropic from federal use, finding the White House and Pentagon likely retaliated against the AI company for speaking out about contract limits, a move the court called unconstitutional.

US District Judge Rita Lin, approved Anthropic’s request for a temporary stop. 

This means that while the lawsuit continues, two government actions are blocked including an order from the President telling government agencies to immediately stop using Anthropic’s AI tools and the Pentagon's decision to label Anthropic a “supply chain risk”, which is a label usually reserved for companies thought to be linked to enemy nations.

The final decision in the case will not happen for several months.

“Punishing Anthropic for bringing public scrutiny to the government’s contracting position is classic illegal First Amendment retaliation,” wrote Judge Lin.

How a contract clause became a constitutional fight

Anthropic signed a US$200 million contract with the Department of Defense (DoD) in July 2025 tied to deploying Claude models on the Pentagon’s GenAI.mil platform.

Talks collapsed over a clause granting DoD broad permission to use Anthropic’s tools for “any lawful use.”

Anthropic, led by CEO Dario Amodei, refused to proceed without explicit guardrails barring use in fully autonomous weapons and mass domestic surveillance.

In late February, Defense Secretary Pete Hegseth labelled Anthropic a supply chain risk, requiring major contractors including Amazon, Microsoft and Palantir to certify they weren’t using Claude in military work.

Days later, President Trump posted on Truth Social ordering agencies to phase out Anthropic tools within six months, alongside the formal White House directive to “immediately cease” use.

Inside the ruling

Judge Lin said statements by President Trump and Secretary Hegseth, which labelled  Anthropic “woke” and its staff “left-wing nut jobs”, had little to do with genuine security concerns.

“Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the US for expressing disagreement with the government,” she wrote.

The Pentagon claimed that Anthropic’s position created a security risk. However, the court pointed out that if this were simply a routine disagreement over a contract, the DoD could have just stopped using the Claude AI tool.

The court found the actions “far exceed” what could reasonably address any stated national security interest.

What it means for agencies, contractors and IT buyers

Agencies and defence contractors can keep using Anthropic’s Claude models while the case proceeds, absent new government action.

Plus, the immediate certification burden tied to the supply-chain designation is paused, easing near-term risk for integrators and cloud partners.

Because the administration relied on different legal authorities, Anthropic is challenging the supply-chain designation separately in the US Court of Appeals in Washington.

The dispute could resolve if the parties reach contract terms on permissible use – or escalate if courts set broader limits on punitive procurement actions tied to speech.

This key timeline outlines the rapid escalation of the dispute, which began with a contract disagreement over AI guardrails and culminated this week with a judge halting the President's and Pentagon’s punitive directives against Anthropic:

• July 2025: Anthropic signs US$200m DoD contract.
• Fall 2025: Talks break down over “any lawful use” clause; Anthropic seeks explicit AI guardrails.
• Late February 2026: DoD labels Anthropic a supply chain risk; primes told to certify non-use.
• Days later: President orders agencies to phase out Anthropic tools; separate directive to “immediately cease” use.
• This week: Judge Lin grants preliminary injunction halting both actions.

Anthropic said it was “grateful to the court for moving swiftly,” adding that while the case protects the company and its partners, its focus remains on working with the government to deliver “safe, reliable AI.”