WASHINGTON, Feb. 25, 2026 — The Pentagon has given Anthropic until Friday to allow unrestricted military use of its AI systems or face action that could include a Defense Production Act order or a supply-chain risk designation after Defense Secretary Pete Hegseth met with CEO Dario Amodei, Reuters reported.
Anthropic has said it will not allow its Claude models to be used for fully autonomous weapons or domestic surveillance of U.S. citizens, framing those limits as ethical red lines even as the company works with national security customers, AP reported.
Guardrails and lawful use
Pentagon officials say military use should be governed by U.S. law and lawful orders rather than by a private contractor’s usage policy, and they want access to Anthropic’s tools across lawful defense use cases, CNBC reported.
Anthropic has said it is in good-faith talks to support the government’s national security mission while keeping safeguards in place that it says its models can reliably and responsibly uphold, Reuters reported.
Pentagon expands AI vendors
The dispute comes as the Pentagon broadens its AI procurement, with defense contracts worth up to $200 million awarded to Anthropic, Google, OpenAI and Elon Musk’s xAI, AP reported.
Reuters reported that Anthropic had been the only large-language-model provider on classified networks until the Pentagon reached an agreement this week to deploy xAI across those systems, a shift that raises the stakes of the standoff, Reuters said.
Pressure tactics and timeline
Invoking the Defense Production Act in an AI guardrails dispute would be an unusual expansion of the law’s use, which has historically compelled industry to meet national defense needs during emergencies, TechCrunch reported.
A supply-chain-risk label would pressure government contractors to avoid Anthropic’s models, a designation typically reserved for foreign adversaries and one that could ripple across federal procurement, CNBC reported.
DW reported that the Pentagon’s Friday deadline remains in place as talks continue, with the outcome expected to shape how far military agencies can push private AI providers to relax safety policies, DW said.
How we report: We select the day’s most important stories, confirm facts across multiple reputable sources, and avoid anonymous sourcing. Our goal is clear, balanced coverage you can trust—because transparency and verification matter for informed readers.
Image Attribution ▾
Image: “The Pentagon US Department of Defense building” by DoD photo by Master Sgt. Ken Hammond, U.S. Air Force. Public domain (U.S. federal government work). Source: https://commons.wikimedia.org/wiki/File:The_Pentagon_US_Department_of_Defense_building.jpg. Modifications: cropped and resized to 1920×1080 (16:9).
