A direct clash between the Pentagon and Anthropic is moving into the open. Defense Secretary Pete Hegseth is calling in Anthropic CEO Dario Amodei to the Pentagon on Tuesday morning to discuss the military use of Claude, according to reporting from Axios.
The meeting comes with unusually high stakes for an AI company already working with the Department of Defense. The Pentagon is threatening to declare Anthropic a “supply chain risk” after the company refused to allow certain uses of its technology.
Why Hegseth is calling Amodei
The immediate issue is the Department of Defense’s access to Claude, Anthropic’s AI technology. Anthropic signed a $200 million contract with DOD last summer, putting the company in a formal relationship with the military.
That relationship is now under pressure. According to the source cited by Axios, Hegseth is giving Amodei an ultimatum: play ball or be banished.
The phrase captures the basic choice now facing Anthropic. The company can allow the Department of Defense to use Claude in the ways being requested, or it can risk losing its place as a Pentagon partner.
What Anthropic refused to allow
The dispute centers on two uses Anthropic has rejected. The company refused to allow the Department of Defense to use its tech for the mass surveillance of Americans and the development of weapons that fire without human involvement.
Those refusals are the reason the Pentagon is considering the “supply chain risk” designation. The label is typically reserved for foreign adversaries, which makes its possible use against Anthropic especially significant.
The source article does not say whether the designation has already been applied. It says the Pentagon threatens to declare Anthropic a “supply chain risk,” and that the meeting is intended to address the military use of Claude.
Why the contract matters
The business stakes are clear because the source article identifies a specific contract and a specific consequence. Anthropic signed a $200 million contract with DOD last summer. A “supply chain risk” designation would void Anthropic’s contract and force other Pentagon partners to drop Claude entirely.
That would make the pressure broader than a single meeting or a single agency relationship. If the designation were applied, it would not only affect the contract between Anthropic and DOD. It would also affect other Pentagon partners using Claude.
The source article also notes that replacing Anthropic would be a significant undertaking. That matters because it suggests the Pentagon’s threat may carry practical costs for both sides. Anthropic faces the risk of losing the contract and related Pentagon work, while the Department of Defense would face the challenge of moving away from Claude.
How Claude’s reported use raised tensions
The tensions did not remain abstract. Claude was reportedly used during the January 3 special operations raid that resulted in the capture of Venezuelan president Nicolás Maduro. The source article says that episode brought the two sides’ tensions into the open.
That reported use gives the dispute a concrete backdrop. Anthropic’s technology was already connected to a military operation, while the company was also refusing to allow other military uses it considered out of bounds.
The result is a conflict over limits. The Department of Defense wants access to AI technology for military purposes. Anthropic is trying to hold lines around mass surveillance of Americans and weapons that fire without human involvement.
What remains unresolved
Several core questions are still open based on the source article. It is unclear whether Hegseth is bluffing. It is also unclear what Amodei will agree to, if anything, during the Pentagon meeting.
What is clear is that the “supply chain risk” threat gives the Pentagon leverage. If used, it would end Anthropic’s DOD contract and require other Pentagon partners to drop Claude.
For Anthropic, the meeting tests whether the company can keep its Department of Defense work while maintaining the restrictions it has placed on certain uses of its technology. For the Pentagon, it tests how far it is willing to go to pressure an AI supplier that has refused some requested military uses.
The outcome has not been reported in the source article. But the stakes described there are direct: Claude’s military use, Anthropic’s $200 million DOD contract, and the possibility that a label typically reserved for foreign adversaries could be turned against a major AI firm.