A view of the Pentagon
A view of the Pentagon. Home to the US Defense Department, the Pentagon is one of the world's largest office buildings. DANIEL SLIM/Getty Images

The Pentagon switched on OpenAI's ChatGPT Mil and Elon Musk's Grok for Government inside its GenAI.mil hub on Monday, giving more than 3 million military and civilian personnel a choice of three frontier AI models — while Anthropic's Claude remained locked out, days after a federal judge ruled the Pentagon's case against the company illegal.

GenAI.mil launched in December with Google's Gemini as its only option and had already signed up more than 1.7 million users before Monday's expansion, the Department of War said. Both new additions cleared Impact Level 5 accreditation, the authorization tier covering controlled unclassified information, and run in isolated government cloud infrastructure separate from the companies' consumer products.

ChatGPT Mil, built through OpenAI's government program, centers on chat, file handling and custom GPTs for document-heavy work such as policy drafting, logistics planning and compliance checklists. Grok for Government, offered through Elon Musk's Starshield AI arm, is pitched for supply-chain tracking and research tasks. "Access to a diverse suite of AI capabilities from across the resilient American technology stack will give warfighters the tools they need to act with confidence," a War Department official told DefenseScoop.

Anthropic's Claude is not part of the rollout. The company had been in talks to join GenAI.mil alongside Gemini but refused Pentagon requests to drop contract terms barring its models from mass domestic surveillance and fully autonomous weapons use. Defense Secretary Pete Hegseth responded by designating Anthropic a national security supply-chain risk, barring defense contractors from using Claude. Anthropic sued, and on Aug. 27 — four days before Monday's launch — U.S. District Judge Rita Lin ruled the designation "illegal and baseless," finding it amounted to unconstitutional retaliation against the company for its public stance. The government is expected to appeal.

That leaves GenAI.mil running with three of the four AI labs the Pentagon originally said would join the platform, and no timeline yet for when — or whether — Claude will be the fourth. The Pentagon has not said whether Monday's launch changes its position now that the court has struck down the risk designation.

Read more