Court rules Pentagon can't blacklist AI firm Anthropic over its ethics rules

A federal judge ruled late Thursday that the Pentagon's ban on Anthropic's AI was "illegal and baseless," foreclosing the use of federal procurement to coerce AI vendors into dropping ethics clauses from their terms of service.

The dispute began in fall 2025, when the Pentagon asked Anthropic to bring its Claude assistant to an internal platform called GenAI.mil. Anthropic's terms bar customers from using Claude for mass surveillance of Americans or to build autonomous weapons; the Pentagon asked Anthropic to replace that clause with language permitting "all lawful uses." Anthropic refused.

President Trump then banned federal agencies from using Anthropic's models, and Defense Secretary Pete Hegseth separately designated the company as a supply-chain risk, blocking defense contractors from using Claude. Anthropic sued in March 2026.

Judge Rita Lin rejected both Pentagon arguments. The first, that Anthropic had "backdoor access" to its technology in national security systems, collapsed after U.S. officials conceded on the docket that no such access exists. The technology, Lin wrote, "is itself no riskier to the national security than any other 'black box' artificial intelligence model."

The second argument, that Anthropic had criticized the administration and acted with "increasingly hostile manners through the press," was constitutionally insufficient. "Neither the Constitution nor the federal statute invoked by Defendants allows them to impose sweeping penalties based principally on Anthropic's critique of the Administration's views," Lin wrote.

The ruling reopens Anthropic as a federal supplier and unwinds the supply-chain-risk designation. It does not settle the underlying policy question of how the Pentagon defines "lawful uses" of commercial AI.