The Pentagon now has its own version of ChatGPT and Grok
TechCrunch Kirsten Korosec
The Pentagon rolled out its own ChatGPT and Grok for 3 million staff. It wants AI speed without sending sensitive data through consumer apps.
Based on reporting by TechCrunch, Kirsten Korosec — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The Pentagon has put ChatGPT Mil and Grok for Government into a secure portal called GenAI.mil, giving civilian and military personnel access to custom versions of two familiar consumer AI tools. The Department of Defense says the system is built for “warfighter needs,” but the day-to-day pitch is simpler: let people use frontier models without pushing sensitive government information through ordinary consumer services.
GenAI.mil is not new. The portal launched last year and first offered Google Gemini. The department says it has already drawn more than 1.7 million unique users out of roughly 3 million personnel, which helps explain why the Pentagon is widening the menu instead of treating this like a pilot that might quietly disappear.
ChatGPT Mil comes through OpenAI for Government and is meant to feel close to the commercial product. The Defense Department says it will center on chat, files, projects, and custom GPTs, with a focus on unclassified work that tends to be heavy on documents: administrative tasks, logistics, planning, and policy. More features are supposed to arrive later.
Grok for Government is being sold in a much more forceful register. The Pentagon says it will improve productivity, knowledge continuity, and collaboration, and it ties the tool to mission work ranging from market research for acquisition staff to supply chain management for logisticians. The branding may be different, but the point is the same: move work faster while keeping the data inside a controlled system.
The rollout also says something about who is in and who is out. Anthropic’s Claude is not part of this setup, after a dispute in which Anthropic resisted giving the Pentagon unrestricted access and wanted safety guardrails instead. The department has meanwhile been cutting deals with other tech companies, including Amazon Web Services, Microsoft, Nvidia, and Reflection AI, which suggests it wants as many AI options as it can get—so long as they come with a lock on the door.
My take — AI-written commentary, not fact-checked reporting
The Pentagon is doing what big institutions always do when a shiny tool shows up: it wants the speed and the control, preferably at the same time. That’s the real story here, not the branding. The awkward part is that the most military-sounding option is also the one marketed with the most confidence, which is very on brand for government AI buying in 2026.
Read more about this at: TechCrunch