Secretary of War Pete Hegseth has given Anthropic CEO Dario Amodei until Friday evening to grant the U.S. military unfettered access to the company's artificial intelligence model, Claude, or face severe consequences, Axios reported.
In a tense meeting Tuesday, Hegseth warned that the Pentagon could either cut ties with Anthropic and designate it a "supply chain risk" or invoke the Defense Production Act to compel the company to tailor its model to military needs.
The ultimatum escalates a growing dispute over AI safeguards and national security.
Pentagon officials acknowledge they are reluctant to lose access to Claude, currently the only AI model used for the military's most sensitive classified work.
"The only reason we're still talking to these people is we need them and we need them now," one official told Axios ahead of the meeting. "The problem for these guys is they are that good."
Anthropic has said it is willing to adapt its usage policies for the Pentagon but will not permit its technology to be used for mass surveillance of Americans or for weapons systems that operate without human oversight.
Accounts of the meeting differed slightly. One senior official described it as "not warm and fuzzy at all," while another source said it remained cordial, with no raised voices and Hegseth praising Claude's capabilities.
Hegseth made clear he would not allow a private company to dictate operational terms or object to specific military use cases.
A key flashpoint involved claims that Anthropic raised concerns with partner Palantir about Claude's use during a raid targeting Venezuelan leader Nicolas Maduro.
Amodei denied raising such concerns and said the company's red lines have not impeded Pentagon operations.
Underscoring the stakes, Hegseth was joined at the meeting by several top officials, including Deputy Secretary Steve Feinberg and the Pentagon's general counsel, Earl Matthews.
The Defense Production Act grants the president authority to compel private companies to prioritize contracts deemed necessary for national defense.
Though used during the COVID-19 pandemic to boost vaccine and ventilator production, it is rarely deployed in an adversarial manner.
War Department officials suggested it could be used to require Anthropic to modify Claude without the safeguards the company insists upon. Anthropic could challenge such a move in court.
Alternatively, the Pentagon could sever its contract and label Anthropic a supply chain risk, forcing defense contractors to certify that Claude is not embedded in their systems.
Such a move would require a replacement for Claude, which is also used in bureaucratic and operational military functions through a partnership with Palantir.
Potential alternatives are emerging.
Elon Musk's xAI recently secured a contract to bring its Grok model into classified environments, though it is unclear whether it could replace Claude.
The Pentagon is also accelerating discussions with OpenAI and Google to move their models into classified systems.
Sources say Claude currently outperforms rivals in some military-relevant applications, including offensive cyber capabilities.
Despite the confrontation, Anthropic struck a conciliatory tone after the meeting, saying Amodei expressed appreciation for the department's work and that discussions were continuing in good faith to support national security within the bounds of responsible AI use.
James Morley III ✉
James Morley III is a writer with more than two decades of experience in entertainment, travel, technology, and science and nature.
© 2026 Newsmax. All rights reserved.