Dario Amodei said Thursday that Anthropic plans to challenge in court the Pentagon’s decision to designate AI companies as supply chain risks, a designation he called “legally unsound.”
The statement came hours after the Pentagon formally designated Anthropic as a supply chain risk after weeks of controversy over how much control the military should have over its AI systems. Supply chain risk designations could prohibit companies from working with the Department of Defense or its contractors. Amodei drew a clear line that Anthropic’s AI would not be used for mass surveillance of Americans or for fully autonomous weapons, but believed the Pentagon should have unrestricted access for “all lawful purposes.”
Amodei said in a statement that the majority of Anthropic’s customers will not be affected by the supply chain risk designation.
“For our customers, this clearly applies only to the customer’s use of Claude as a direct part of a contract with the Department of the Army, and does not apply to all use of Claude by customers with such contracts,” he said.
In a precursor to what Anthropic will argue in court, Amodei said the department’s letter classifying the company as a supply chain risk was narrow in scope.
“This law exists to protect the government, not to punish suppliers. In fact, this law requires the Secretary of the Army to use the least restrictive measures necessary to achieve the goal of protecting the supply chain,” Amodei said. “Even if you are a Department of the Army contractor, the supply chain risk designation does not (and cannot) limit your use of Claude or your business relationship with Anthropic Corporation if it is unrelated to your specific Department of the Army contract.”
Amodei reiterated that Anthropic had been having productive conversations with the Pentagon over the past few days, but said some suspects may have derailed the conversation after an internal memo Anthropic sent to staff was leaked. In it, Amodei characterized the deal between rival OpenAI and the Pentagon as “safety theater.”
tech crunch event
San Francisco, California
|
October 13-15, 2026
OpenAI has signed a contract to work with the Department of Defense on behalf of Anthropic, a move that has sparked a backlash among OpenAI staff.
Amodei apologized for the leak in a statement Thursday and insisted the company did not intentionally share the memo or direct anyone to share it. “It is not in our interest to escalate the situation,” he said.
Amodei said the memo was written “within hours” of a series of announcements, including the president’s Truth Social post to remove Anthropic from the federal system, then Secretary of Defense Pete Hegseth’s supply chain risk designation, and finally the announcement of the Pentagon’s agreement with OpenAI. He called it a “difficult day for the company,” apologized for his tone and said the memo did not reflect his “careful or considered views.” It was written six days ago and is now an “outdated assessment,” he added.
He concluded by saying that Anthropic’s top priority is to ensure that U.S. soldiers and national security professionals maintain access to critical tools during ongoing large-scale combat operations. Anthropic currently supports some of the U.S. operations in Iran, and Amodei said the company will continue to provide models to the Department of Defense at a “nominal cost” “for as long as necessary for the transition.”
Anthropic could likely challenge the designation in federal court in Washington, but the law behind the decision limits the usual ways for companies to challenge government procurement decisions and gives the Pentagon broad discretion on national security issues, making the challenge difficult.
Or, as Dean Ball, a former White House AI adviser during the Trump administration who spoke out against Hegseth’s humane treatment, said: “Courts are pretty reluctant to second-guess the government about what is and isn’t a national security issue…There’s a very high bar to clear to do that. But it’s not impossible.”
