Anthropic challenges Department of Defense supply chain mark in court

Dario Amodei said Thursday that Anthropic plans to challenge the Defense Department’s decision to classify the artificial intelligence company as a supply chain risk in court, a designation he called “legally unsound.”

The statement comes just hours after the department officially designated Anthropic as a supply chain risk following a weeks-long dispute over how much control the military should have over artificial intelligence systems. Identifying supply chain risks could prevent a company from working with the Pentagon and its contractors. Amodei has drawn a firm position that Anthropic’s AI should not be used for mass surveillance of Americans or in fully autonomous weapons, but the Pentagon believes it should have unfettered access “for all lawful purposes.”

Amodei said in his statement that the vast majority of Anthropic’s customers are not affected by the supply chain risk rating.

“As far as our customers are concerned, this obviously only applies to customers’ use of Claude As a direct part of Contracts with the Ministry of War, and not all use of Claude by customers who have such contracts.

As a preview of what Anthropic is likely to argue in court, Amodei said the department’s letter describing the company as a supply chain risk is narrow in scope.

“It exists to protect the government, not to punish the supplier; in fact, the law requires the Secretary of War to use it Less restrictive means are necessary “To achieve the goal of protecting the supply chain,” Amodei said, “even for War Department contractors, identifying supply chain risks does not (and cannot) limit Claude’s uses or business relationships with Anthropic if those are not tied to their specific War Department contracts.”

Amodei reiterated that Anthropic had been in productive talks with the ministry over the past few days, talks that some suspected were derailed when an internal memo he sent to staff was leaked. In it, Amdoudi described rival OpenAI’s dealings with the Department of Defense as “security theater.”

OpenAI signed a deal to work with the Department of Defense instead of Anthropic, a move that sparked backlash among OpenAI employees.

Amodei apologized for the leak in his statement on Thursday, claiming that the company did not intentionally share the memo or direct anyone else to do so. He added: “It is not in our interest to escalate the situation.”

Amodei said the memo was written within “a few hours” of a series of announcements, including a presidential Social Truth circular saying Anthropology would be removed from federal systems, then Defense Secretary Hegseth’s supply chain risk outlining, and finally the announcement of the Pentagon’s deal with OpenAI. He apologized for his tone, calling it a “difficult day for the company” and saying the memo did not reflect his “accurate or considered views.” He added that it was written six days ago and is now an “outdated assessment.”

He concluded by saying that Anthropic’s top priority is ensuring that U.S. soldiers and national security experts maintain access to critical tools in the midst of major ongoing combat operations. Anthropic currently supports some US operations in Iran, and Amodei said the company will continue to provide its models to the Department of Defense at a “nominal cost” “for as long as necessary to make this transition.”

Anthropic could appeal the decision in federal court, most likely in Washington, but the law behind the decision makes it difficult to appeal because it limits the usual ways companies can appeal government procurement decisions and gives the Pentagon broad discretion over national security matters.

Or as Dean Paul, a former White House AI adviser under Trump who has spoken out against Hegseth’s treatment of anthropology, put it: “Courts are largely reluctant to second-guess the government about what is or is not a national security issue… There’s a very high bar one needs to cross in order to do that. But it’s not impossible.”