by Akash Sriram
March 9 (Reuters) – Anthropic sued the U.S. government on Monday, escalating a dispute that the artificial intelligence company sees as retaliation for refusing to lift safety restrictions on its Claude model.
The Amazon-backed company said it was willing to work with the military. Just without any conditions.
It also filed a related case with the U.S. Court of Appeals for the D.C. Circuit challenging the separate legal authority cited by the government.
The following narrative is based on the allegations made by Anthropic in its lawsuit.
What is the controversy Anthropic is talking about?
Anthropic said it spent years making Claude the most widely deployed cutting-edge AI model in government, including on classified military networks, developing a specialized “Claude Gov” version and relaxing many of the standard restrictions to suit national security efforts.
The conflict began during the Pentagon’s GenAI.mil platform negotiations in the fall of 2025, when the Department of Defense demanded that Anthropic abandon its usage policy entirely and allow Crowder to be used for (in the government’s words) “all lawful uses.”
Anthropic said it largely agreed, except for two points it considered non-negotiable: It would not allow Cloud to be used in lethal autonomous warfare or for mass surveillance of Americans without human oversight.
The company says Cloud has not been tested for these uses and is not safe for them. The company said it was also willing to help move the work to another provider if a deal couldn’t be reached.
Pentagon officials have offered varying accounts of how the dispute began. The unit’s chief technology officer said publicly that tensions rose after the U.S. raid in Venezuela, when an Anthropic executive called his Palantir counterpart to ask if Crowder was involved in the operation.
That account does not appear in Anthropic’s complaint.
From ultimatum to total ban
Defense Secretary Pete Hegseth met with Anthropic CEO Dario Amodei on February 24 with an ultimatum: Comply within four days or face one of two penalties: enforcement under the Defense Production Act or expulsion from the defense supply chain on the grounds of “national security risk.”
Amodei publicly rejected the request on February 26. The next day, ahead of a 5:01 PM ET deadline, President Donald Trump issued a directive on Truth Social ordering all federal agencies to immediately stop using Anthropic’s technology.
In social media posts, the president described Anthropic as a “radical left-wing, woke company.”
Hours later, Hegseth declared on X that Anthropic was a “supply chain risk to national security” and that no military contractor or supplier could do commercial business with the company.
Agencies quickly lined up. The U.S. General Services Administration terminated Anthropic’s government-wide contract. The Treasury Department, state governments and the Federal Housing Finance Agency publicly severed ties. Anthropic’s complaint alleges that the Pentagon used Anthropic’s tools to launch a massive airstrike against Iran just hours after the ban was implemented.
White House spokesperson Liz Houston said the government will not allow a company to “endanger our national security by dictating how the world’s most powerful military operates,” adding that the U.S. military “will never be hijacked by the ideological whims of Big Tech leaders” and will abide by the Constitution and “will not abide by any artificial intelligence company’s terms of service.”
Why Anthropic decided to sue
Anthropic believes that the supply chain designation has no factual basis. The company pointed to its FedRAMP authorization, positive security review and years of government praise, including from Hegseth, who called Claude’s abilities “superb” during a Feb. 24 meeting.
Two senior Pentagon officials later told reporters that there was “no evidence of a supply chain risk” and that the designation was “ideologically driven.”
Anthropic brought five legal claims, arguing that these actions violated the Administrative Procedure Act, the First Amendment, the Fifth Amendment, the President’s statutory powers, and the APA’s prohibition on unauthorized agency sanctions.
(Reporting by Akash Sriram in Bengaluru)