Pentagon Taps OpenAI, Bans Anthropic
In a major shakeup for the AI industry, the Trump administration has ordered all federal agencies to stop using technology from AI startup Anthropic, citing "supply chain risk." Just hours later, rival OpenAI announced a major partnership with the Pentagon, positioning itself as a core AI supplier for the U.S. government and military.
The rift between Anthropic and the Pentagon escalated over several months, culminating in a firm deadline from Defense Secretary Pete Hegseth. Anthropic was given until 5:01 p.m. on February 27, 2026, to agree to the military's terms or face termination of its $200 million contract and designation as a "supply chain risk." The core of the dispute was Anthropic's refusal to remove safeguards preventing its AI model, Claude, from being used for mass domestic surveillance and in fully autonomous weapons systems without human oversight. The Pentagon insisted on the right to use the technology for "all lawful purposes," a stance Anthropic argued was too broad and could override its ethical commitments. The "supply chain risk" designation is a significant move, historically applied to foreign adversaries, not American companies. This action could prevent any contractor or supplier doing business with the U.S. military from also working with Anthropic, potentially impacting its partnerships with major tech firms like Amazon, Microsoft, and Nvidia that are also key military suppliers. Anthropic has stated it will challenge the designation in court. In contrast, OpenAI's new partnership falls under its "OpenAI for Government" initiative, launched in June 2025. This program consolidates its work with various U.S. agencies, including NASA, the Treasury Department, and the Air Force Research Laboratory, offering access to its most advanced models in secure environments. While OpenAI's deal with the Pentagon includes similar "red lines" to what Anthropic requested—prohibiting mass surveillance and requiring human responsibility for the use of force—it was framed within the "all lawful uses" clause the Pentagon demanded. OpenAI CEO Sam Altman stated the Pentagon agrees with these principles and they were incorporated into their agreement. The ban affects more than just the Pentagon; Anthropic's AI, Claude, was already being used or tested in at least five federal agencies, including the Departments of Energy and Homeland Security. These agencies now have a six-month window to phase out their use of Anthropic's technology. The decision has drawn sharp reactions, with political battle lines forming in Silicon Valley. Some tech leaders have pledged their companies would not question Pentagon policies, while many AI researchers and employees at rival companies have publicly supported Anthropic's stand. Anthropic was the first AI company to have its model deployed on the Pentagon's classified networks. OpenAI, despite its new agreement, did not have the necessary infrastructure for classified systems until a recent major investment from Amazon, whose cloud services are used by the Pentagon.
