Pentagon rolls out ChatGPT Mil and Grok for Government to 3 million staff

Pentagon adds ChatGPT Mil and Grok for Government

The US Department of Defense now runs its own versions of two of the best-known consumer AI assistants. According to TechCrunch, custom builds of OpenAI's ChatGPT and xAI's Grok have been deployed across the department, putting generative AI in the hands of roughly 3 million military and civilian personnel, with the tools adapted to what the department calls "warfighter needs."

The two products, named ChatGPT Mil and Grok for Government, live inside GenAI.mil, a centralized portal launched last year. When the portal debuted it carried Google's Gemini; its purpose is to let defense staff use commercial frontier models without routing sensitive government information through ordinary consumer services. The military versions also avoid the data collection that consumer products typically perform.

Take-up has been quick. The department says more than 1.7 million unique users have already signed up out of the 3 million personnel eligible.

Two models, two missions

ChatGPT Mil grew out of OpenAI's OpenAI for Government program and is meant to feel familiar to anyone who has used the commercial chatbot. The department says it supports chat, files, projects and custom GPTs, and is aimed at routine unclassified work such as administration, logistics, planning and policy, with additional features to follow.

Grok has been cast in noticeably more operational terms. In a press release quoted by TechCrunch, the department said Grok for Government, delivered through Starshield AI — a secure satellite network built on SpaceX's Starlink technology — "will provide the warfighter with immediate productivity gains, stronger knowledge continuity and more secure and efficient collaboration." The department added that Grok would help the military "execute missions faster and with greater precision," citing use cases that range from market research for acquisition staff to supply chain management for logisticians.

Anthropic is conspicuously absent

The launch also highlights which vendors did not make the cut. Anthropic's Claude is missing from the portal, and TechCrunch links that to a dispute between the Pentagon and the frontier lab. After Anthropic declined to give the military unrestricted use of its models and insisted on certain safety guardrails instead, the Trump administration labeled the company a supply-chain risk — a designation Anthropic is now fighting in court. The Pentagon has since deepened ties with other suppliers, including Amazon Web Services, Microsoft, Nvidia and Reflection AI.

Why it matters

At 3 million potential users, this is one of the largest institutional rollouts of generative AI anywhere, and a marker of how quickly the technology has moved from novelty to standard government infrastructure. The architecture matters too: by wrapping commercial models in a dedicated, secure portal, the department gets frontier capabilities without the privacy exposure that comes with consumer products, and it has set a template other agencies are likely to copy.

The vendor mix is arguably the more consequential detail. Anthropic's exclusion — tied to its insistence on usage limits — suggests that a company's willingness to grant unrestricted military use may now determine whether it wins large government contracts. That is a precedent the rest of the AI industry, and other governments watching from the sidelines, will not ignore.