Sen. Elizabeth Warren is pressing the Pentagon over its decision to designate Anthropic a supply-chain risk, calling the move “retaliation” ahead of a key federal court hearing in San Francisco.
Warren sent her March 23 letter to Defense Secretary Pete Hegseth one day before US District Judge Rita Lin is set to hear Anthropic’s request for a preliminary injunction that would pause the designation while the company’s First Amendment lawsuit moves forward.
Warren’s letter puts the Pentagon’s rationale under pressure
In a March 23 letter to Hegseth, Warren called the designation “retaliation” and argued that the Defense Department could have ended its contract with Anthropic instead of using a supply-chain risk tool to cut the company off more broadly, according to TechCrunch.
That dispute is now at the center of the case. Anthropic says the government targeted it after the company refused to allow two uses: mass surveillance of Americans and fully autonomous weapons capable of lethal action without human intervention. The Pentagon says the designation reflects a national security judgment.
Earlier this month, the Pentagon formally designated Anthropic a supply-chain risk, making it the first known public use of that label against a US company. That raises the stakes well beyond a single contract fight, especially given the Claude AI model’s broader commercial applications beyond government work.
Court filings are testing the Pentagon’s case
Recent filings have added pressure to the Pentagon’s timeline. TechCrunch reported that on March 4, one day after the designation was finalized, Under Secretary Emil Michael emailed Anthropic CEO Dario Amodei to say the two sides were “very close” on the same issues the government now cites as evidence of national security risk.
Anthropic has also pushed back on one of the Pentagon’s central claims: that the company could disable or alter Claude during military operations. In sworn declarations cited by TechCrunch, Anthropic said that once Claude is deployed in an air-gapped, classified environment operated by a third-party contractor, the company has no remote access, no back door, and no kill switch.
Reuters added another complication, reporting that the Pentagon has continued relying on Anthropic’s technology for military operations, including work tied to Iran, according to a source familiar with the matter. That does not settle the legal dispute, but it does complicate the government’s claim that Anthropic poses an unacceptable security risk.
The hearing could shape more than Anthropic’s immediate position with the Pentagon. If the designation stays in place while the case proceeds, AI companies negotiating with the government may have to weigh whether safety limits on military use can trigger a blacklisting fight.
Questions about supply chains and security are already hitting other corners of the AI market, from Elon Musk’s reported $2.9 billion purchase of solar equipment from China to a federal case over a $2.5 billion scheme to ship Nvidia AI servers to China.
Also read: Anthropic’s new institute is another sign that AI governance is moving from theory into product and policy fights.