A federal appeals court in Washington rejected Anthropic PBC’s bid to block, for now, the Pentagon’s “supply chain risk” designation while the case proceeds.

Anthropic, an AI developer based in San Francisco and best known for its Claude chatbot, said it disagreed and was considering further review.

The Pentagon argued that limits Anthropic placed on Claude, including bans on lethal autonomous warfare and domestic surveillance, created a national security risk because they could restrict uses the Pentagon considers legitimate.

Under Pentagon authority to exclude sources from some procurements or require removals from contracts, the dispute could have commercial implications for Anthropic, which announced a US$200 million Defense Department AI agreement in 2025.

A separate injunction issued in San Francisco in March remains in effect and blocks a broader government-wide ban on new contracts involving Anthropic, making the latest ruling narrower than a government-wide prohibition.

The case could be a test case for whether procurement tools can be used against AI vendors that maintain restrictions on military use.

🔗 Source: Bloomberg

Recent Anthropic developments