A federal judge has vacated the Pentagon’s designation of Anthropic as a national-security supply-chain risk, ruling that the government imposed broad penalties in unconstitutional retaliation for the AI company’s restrictions on military use.
US district judge Rita Lin barred enforcement of the February 27 designation and removed sanctions imposed by nine federal agencies. Her decision does not require the Department of Defense to buy or use Anthropic’s models, and the department is expected to appeal.
The distinction leaves the government free to choose another vendor while limiting its ability to punish Anthropic across the wider federal contracting system.
The ruling separates procurement choice from retaliation
Defense secretary Pete Hegseth’s designation made Anthropic ineligible for federal contracts. An additional measure prevented military contractors and suppliers from doing business with the company, extending the consequences beyond the Pentagon’s own purchases.
Lin described those measures as arbitrary, capricious and contrary to law. According to Wired, she also found that the Pentagon, Treasury, State Department, Department of Homeland Security and five other agencies had improperly imposed sanctions. Her ruling removes those penalties.
At the same time, Lin affirmed that the Pentagon is not required to deploy Anthropic technology. The department can select the AI vendor it prefers and can stop using Claude without blacklisting the company throughout federal supply chains.
That boundary is central to the decision. It does not grant Anthropic a military contract. It constrains the government’s response to a vendor disagreement by distinguishing a purchasing decision from a broad national-security designation.
Military-use limits triggered the dispute
The case grew from a $200 million agreement involving Claude models for military applications. Anthropic sought restrictions on support for lethal autonomous weapons and mass-surveillance systems. Hegseth rejected those limits and argued that a contractor could not dictate how its technology was used once deployed under a contract permitting all lawful use.
Tensions increased after reports that the United States had used Claude during the operation to capture Venezuelan president Nicolas Maduro. Wired says a Palantir employee relayed concerns from an Anthropic staff member to US officials about the model’s use.
When negotiations collapsed, Hegseth applied the supply-chain-risk label. The Pentagon said Anthropic’s ability to disable or alter its technology could create unacceptable danger if the models were connected to classified systems during a war.
Anthropic responded with lawsuits in California federal court and the US court of appeals for the District of Columbia, alleging violations of its First and Fifth Amendment rights. The California ruling addresses the designation and related sanctions; the separate DC case remains active.
Evidence undercut the government’s security rationale
Lin found that the government’s own conduct conflicted with its claim that Anthropic posed a genuine sabotage risk. Her ruling noted that officials were still discussing collaboration with the company on a new model called Mythos in sensitive contexts.
The judge treated those conversations as inconsistent with the idea that Anthropic might poison its software to harm national security. The evidence supported her conclusion that the measures were retaliatory rather than a justified response to an established supply-chain threat.
Anthropic welcomed the ruling and said it remained focused on working productively with the government on national security. The Pentagon had not provided Wired with an immediate comment, and an appeal was expected.
Several limits remain. The decision does not settle Anthropic’s separate appellate case, resolve the policy conflict over autonomous weapons and surveillance, or guarantee future defense work. It also does not prevent agencies from assessing concrete technical risks through lawful procedures.
What it changes immediately is the reach of the punishment. The Pentagon may refuse Claude, but the court found that disagreement over use restrictions did not justify excluding Anthropic across government and forcing military suppliers to sever commercial ties with the company.
