Intercept FOIA Reveals $200M Pentagon AI Deals With Four Labs

Through a Freedom of Information Act lawsuit conducted with Legal Advocates for Safe Science and Technology, The Intercept obtained more than 400 pages of documents on U.S. military contracts with Anthropic, Google, OpenAI, and xAI. Each contract is worth up to $200 million. The four were originally signed in July 2025, and commit the companies to prototype AI tools that "improve military advantage, military utility, or enhance military decision making" across the whole of the armed forces.

The paperwork spells out what collaboration means. Companies will share "benchmarks datasets for DoD use cases" and receive "appropriate briefings on DoD operational missions and threats." Engineers and policy experts from all four contractors will conduct "tabletop exercises," gamified simulations of real-world scenarios, with the Pentagon. Those workers, one document notes, "can be deployed to warfighting support settings including, but not limited to, combatant commands, service components, and theatre components."

OpenAI CEO Sam Altman described the deal as "definitely rushed." Pentagon spokesperson Sean Parnell said, "We must equip our warfighters with 21st-century technology in order to defeat 21st-century threats." OpenAI's Nate Evans said the company's tools "are built to support legitimate national security work…while refusing uses that cross the red lines we publicly announced."

Two of the documents indicate the Department of Defense asked OpenAI to minimize how often its large language model would refuse what was requested by the military. DoD and OpenAI both say those pages were drafts "incorrectly provided to The Intercept."

"This appears to be the first time the extent of collaboration between frontier AI labs and the Pentagon is spelled out in explicit terms," Sophia Goodfriend, a University of Cambridge research fellow, told The Intercept. "What's particularly notable is the terms by which engineers are working in lockstep with the department of war to engineer AI systems for surveillance, targeting, and killing."

Anthropic refused to sign a follow-up deal to allow deployment of its technology on classified military networks without contractual prohibitions against use for domestic spying and autonomous weapons. Secretary of Defense Pete Hegseth then designated the firm a "supply chain risk" in March and banned its services from government use; a federal judge overturned that decision last month. Also in March, the Wall Street Journal reported that despite Hegseth's ban, U.S. Central Command used Anthropic's large language model services to execute its bombardment of Iran, including for "target identification." It is the most consequential thread in our ongoing Anthropic coverage this summer.