Pentagon building government-owned AI to replace Anthropic, officials say
Mar 18th 2026
After contract talks broke down over use limits, the Defense Department is engineering its own large language models for classified use while Anthropic contests a supply-chain designation.
- The Pentagon is developing multiple government-owned large language models for classified environments, with engineering work already underway, according to DOD official Cameron Stanley.
- Anthropic’s proposed $200 million contract with the Department of Defense collapsed after the parties failed to agree on limits on military uses such as mass surveillance and autonomous weapons.
- OpenAI and xAI later reached separate agreements to provide AI services to the Pentagon for classified systems.
- Defense Secretary Pete Hegseth designated Anthropic a supply-chain risk, blocking Pentagon contractors from working with the company, and Anthropic is challenging that designation in court.
- The department says it expects the new government-owned models to be available for operational use very soon, signaling preparations to operate without Anthropic
Articles
- The Pentagon is backing nuclear waste recycling for long-lasting military power sources www.scientificamerican.com
- The Pentagon is developing alternatives to Anthropic, report says techcrunch.com
- The Pentagon Provided a Rare Inside Look at Palantir's Project Maven and How the AI Tool Helps the Military Wage War — “Left click, right click, left click…We've gone from identifying the target, to now coming up with a course of action, to now actioning that target all within one system.” www.businessinsider.com