Anthropic took its fight with the Trump administration to court on Monday, opening a fresh front in one of the ugliest battles in the AI business. The company sued after the administration labeled it a security threat and moved to cut off its federal contracts.
That decision put Anthropic in a category usually linked to hostile foreign players, not a U.S. company building AI models for both government and commercial work.
In its complaint, filed in the Northern District of California, Anthropic argued that the administration acted outside the law and used federal power as punishment after the company pushed back on how the Pentagon wanted to use AI.
The lawsuit named the Defense Department, Defense Secretary Pete Hegseth, Secretary of the Treasury Scott Bessent, Secretary of State Marco Rubio and Secretary of Commerce Howard Lutnick.
Anthropic told the court that the government’s actions threaten one of the fastest-growing private AI companies in the country and could set a dangerous example for other businesses that disagree with Washington. The company asked the court to rule that the moves were unlawful.
The White House hit back fast. A spokeswoman said, “President Trump will never allow a radical-left, woke company to jeopardize our national security by dictating how the greatest and most powerful military in the world operates.”
Researchers back Anthropic after the lawsuit widens the fight across Silicon Valley
Not long after the case was filed, 37 AI researchers from rivals OpenAI and Google submitted a brief asking the court to side with Anthropic. That support showed how far this clash has spread beyond one company and one contract.
Their filing warned that punishing a leading U.S. AI firm over safety limits could hurt the country’s wider position in artificial intelligence.
The researchers wrote, “If allowed to proceed, this effort to punish one of the leading U.S. AI companies will undoubtedly have consequences for the United States’ industrial and scientific competitiveness in the field of artificial intelligence and beyond.”
That brief added more pressure to a case that was already drawing attention across the tech sector.
The deeper fight centers on what rules should exist when the Pentagon uses AI systems. During contract talks with the Defense Department, Anthropic wanted clear guarantees that its tools would not be used for mass domestic surveillance or autonomous weapons.
The Pentagon rejected that approach. Its position was simple: it follows the law, it would not do those things, and the company should trust the military to use AI in any lawful situation. That disagreement helped blow up formal negotiations, which the Pentagon has since said are over.
The fight also spread into politics and trade. The two sides have clashed over Trump’s decision to allow AI chips to be exported to China. There has also been friction over Anthropic’s links to organizations that donated to Democratic causes.
Those issues turned the company into a major target for Trump allies, even as the dispute brought it more support from some customers and partners.
Trump and Hegseth press the crackdown as Anthropic fights to protect a $200 million contract
The clash got much worse on February 27, when Hegseth said he would designate Anthropic a supply-chain risk for the Pentagon. That tool is normally used for companies tied to foreign adversaries.
Under that process, top Pentagon officials must show that a real security threat exists. Hegseth and other officials argued that Anthropic’s refusal to let the military use its AI in all lawful cases was itself a risk.
Their argument was that a private company should not be able to control how the armed forces use critical technology, because a firm could later switch off access or change settings during operations.
That same day, Trump ordered federal agencies to stop using Claude and gave them six months to move to other AI models.Anthropic seized on that point in its complaint, saying the six-month window shows how important its systems are to the government.
The company also said Trump skipped the proper legal steps required to cancel a federal contract. Its Defense Department deal was worth up to $200 million.
The financial damage could reach beyond direct government work. Customers that also deal with the Pentagon may now have to prove they did not use Claude in Defense Department activity.
That could hit Anthropic’s business even outside the contract itself. Still, Microsoft and Google, both investors or partners, said they would keep working with the company on commercial projects that do not involve the Pentagon.
Supporters of Anthropic say the administration’s case looks shaky for another reason: the Pentagon has used Claude in Iran operations, and until recently Anthropic was the only AI model developer cleared for classified settings.
An Anthropic spokeswoman said, “Seeking judicial review does not change our longstanding commitment to harnessing AI to protect our national security, but this is a necessary step to protect our business, our customers and our partners.” She added, “We will continue to pursue every path toward resolution, including dialogue with the government.”
