Anthropic's Pentagon contract blacklisting

The DC Circuit panel that issued a 2-1 ruling upholding the Pentagon's blacklisting of Anthropic has stayed its own ruling pending appeal, and Anthropic said it is "considering all options, including further review." The September 25 ruling held that Claude's built-in safety restrictions can qualify as a supply-chain risk under the Supply Chain Security Act's §4713 because they could cause the model to refuse government-requested tasks, creating operational uncertainty. Defense Secretary Pete Hegseth did not exceed his statutory or constitutional authority in excluding Anthropic from defense contracts, the majority found. Despite the designation, Anthropic's valuation has reportedly grown from roughly $500 billion to roughly $2 trillion and revenue has risen, which one publication argues limits the White House's practical leverage over the company.

The contract dispute originated when Anthropic refused to accept the Pentagon's 'all lawful use' standard for AI deployment, a position other model-makers accepted. Anthropic drew red lines against use of its tools for mass domestic surveillance or autonomous weapons. President Trump gave the Pentagon six months, until August, to remove Anthropic from defense systems. A parallel California proceeding produced a distinct result: Judge Rita Lin issued a preliminary injunction against the Presidential Directive, the Hegseth Directive, and the §3252 supply-chain designation, finding Anthropic likely to succeed on First Amendment retaliation, Fifth Amendment due process, and APA violations. Lin also found the government relied on a risk assessment from the wrong official and had not considered less intrusive measures. The DC Circuit said it had 'no quarrel' with Lin's conclusion that Anthropic lacked the bad motive required under the narrower §3252 standard, while noting §4713 carries a substantially broader definition of supply-chain risk. The two rulings can coexist because they concern different statutory powers.