Unlock the White House Watch newsletter for free
A guide to what President Trump’s second term means for Washington, business, and the world.
The US government has threatened to terminate its contract with Anthropic if the AI startup fails to reach an agreement with the Pentagon, with one senior official accusing Chief Executive Dario Amodei of having a “God complex” after rejecting a “final offer” to continue working with the military.
A government official told the FT on Thursday that a contract signed in August to provide Anthropic’s Claude chatbot to all three government departments would be scrapped unless the startup agreed to give the Pentagon unrestricted use of its technology.
Late Thursday evening, Emile Michael, the undersecretary of defense for research and technology, who is participating in negotiations with Anthropic, called Mr. Amodei a liar with a “god complex.”
“He wants nothing more than to try to personally control the U.S. military and is willing to put the security of our country at risk,” Michael wrote to X.
that’s a shame @DarioAmodei He is a liar and has a god complex. He wants nothing more than to try to personally control the U.S. military, and he is comfortable putting our nation’s security at risk.
of @DeptofWar We always abide by the law, but we don’t bow to the whims of commercial technology… https://t.co/ZfwXG36Wvl
— Undersecretary of the Army Emile Michael (@USWREMichael) February 27, 2026
The escalating controversy is likely to lead to a legal battle between one of the nation’s leading AI research institutes and the Trump administration.
The makers of the Claude model have rejected the Pentagon’s full use of the technology, citing concerns about lethal autonomous weapons and domestic mass surveillance.
Secretary of Defense Pete Hegseth summoned Mr. Amodei to Washington on Tuesday and demanded that Anthropic not allow legal use of its models by Friday or risk being cut from the Pentagon’s supply chain or having its technology stolen.
Pentagon officials sent a final proposal to Anthropic in a letter Wednesday, according to people familiar with the matter. But Amodei said those conditions were unacceptable for a company that bills itself as more responsible and safety-oriented than its competitors.
“These threats do not change our position. We cannot in good conscience comply with their demands,” Amodei said in a blog post Thursday.
“We believe that in limited cases, AI could undermine rather than protect democratic values,” he added. “Some applications are simply beyond what can be done safely and reliably with today’s technology.”
Administration officials told the FT that the government considers Claude a “great product” and hopes a resolution can be reached. However, it added that if this does not happen, chatbots will no longer be available to federal agencies through the General Services Administration.
Amodei said Anthropic remains hopeful about the agreement. “Given the tremendous value that Anthropic’s technology brings to our military, we hope they reconsider,” he wrote.
The Department of Defense did not respond to requests for comment. Sean Parnell, the military’s chief spokesperson, wrote in X earlier Thursday that the military “has no interest in using AI to conduct mass surveillance of American citizens (which is illegal), nor does it want to use AI to develop autonomous weapons that operate without human involvement.”
“We will not let any company dictate the terms of how we make business decisions,” he added.
If the government follows through with the ultimatum, Anthropic (partnered with Palantir and the only AI lab whose models are used for sensitive Department of Defense work) could be cut from the U.S. defense supply chain and lose a $200 million contract.
The supply chain risk designation could trigger a legal challenge from the $380 billion company, according to people familiar with the matter. Companies that have traditionally been considered supply chain risks under U.S. law, such as China’s Huawei, have historically come from hostile countries.
Alan Rosenstein, an associate professor at the University of Minnesota Law School and author of books on AI and the military, said the attack on Anthropic was “probably well outside the scope of the law,” and said he believed the company had “strong legal defenses in the event it was identified as a supply chain risk.”
Hegseth also threatened to invoke the Defense Production Act, a Cold War-era measure that allows presidents to control domestic industries in the national interest.
The law, used by Joe Biden and Donald Trump to ramp up production of medical supplies during the coronavirus pandemic, would allow the Pentagon to use Anthropic’s tools without a contract agreement.
