Anthropic and the Pentagon are in a contract dispute over how the U.S. military can use the company’s AI systems, including Claude. The standoff escalated on Feb. 24, 2026, when Defense Secretary Pete Hegseth delivered an ultimatum to Anthropic CEO Dario Amodei: agree by 5:01 p.m. ET on Feb. 27, 2026, to allow the technology to be used for any lawful military purpose, or face potential penalties that could cut Anthropic out of Defense Department work.
The dispute centers on whether a private AI company’s usage policies can limit military operations that are otherwise legal under U.S. law.
Anthropic has said it wants to support national security missions while keeping guardrails that bar mass domestic surveillance and direct use in lethal autonomous weapons.
What the Pentagon Is Demanding and What Anthropic Has Offered
In earlier negotiations, Anthropic agreed to contract language that would allow the U.S. government to use its AI systems for missile defense and cyber defense, based on accounts of private talks described by people familiar with the matter. Anthropic has said that multiple versions of its proposed contract language would enable support for missile defense and similar uses.
Pentagon leaders have pushed for broader terms. The Defense Department’s position, as described in recent reporting and statements attributed to senior officials, is that the military should be able to deploy commercial AI models for any lawful purpose without being constrained by an individual company’s acceptable-use policy.
Anthropic’s position is that it can support the government’s mission while maintaining restrictions meant to prevent misuse, particularly in areas involving domestic surveillance and fully autonomous lethal decision-making. The company has said it is continuing good-faith discussions about how to align policy with what its models can reliably and responsibly do.
The gap between the two sides is not primarily about whether the Pentagon can use Claude for defensive missions like cyber defense or missile defense. The dispute is over whether Anthropic must remove policy limits that the Pentagon sees as operational constraints and that Anthropic sees as safety and civil-liberties protections.
Why Missile Defense Scenarios Became a Flash Point
According to accounts from Defense Department officials, one point of tension involved hypothetical scenarios, including how AI systems might be used if an adversary launched an intercontinental ballistic missile at the United States.
Pentagon representatives raised concerns that product guardrails could interfere with time-sensitive operations.
Anthropic has disputed the idea that the Pentagon would have to call the company to lift restrictions during each missile defense operation. The company has described that characterization as false.
Even with assurances, Pentagon leadership has signaled it does not want to be dependent on a private firm to adjust access or policies during high-stakes scenarios. That concern has become part of the case for demanding broader, standardized contract terms across vendors.
