The Department of War flipped the switch Monday on two new artificial intelligence tools for its sprawling GenAI.mil platform, giving more than 3 million military and civilian personnel access to OpenAI's ChatGPT Mil and xAI's Grok for Government alongside the Google Gemini tool that started it all. The rollout marks the most significant expansion yet of a platform that has grown from a single-vendor pilot into a multi-model marketplace for unclassified defense work.
ChatGPT Mil includes chat, files, projects, and custom GPTs, and it can help personnel draft and review procurement materials, summarize policy and guidance documents, and generate internal reports and compliance checklists, according to the Dallas Express. The tool supports unclassified work spanning planning, policy, logistics, and administration, and it holds accreditation for handling Controlled Unclassified Information at Impact Level 5. Per OpenAI, the outlet reports, ChatGPT Mil runs inside authorized government cloud infrastructure, its data is isolated from OpenAI's public and commercial models, and the company does not use GenAI.mil data to train or improve those consumer products.
The Department of War and OpenAI began an enterprise partnership in 2025, and Monday's launch is framed as support for the White House AI Action Plan. Beyond the paperwork use cases the Dallas Express outlined, defense officials told Defense One that operational units including U.S. Northern Command plan to use ChatGPT Mil to coordinate disaster response logistics with FEMA, while Joint Staff logisticians intend to use it to reconcile complex paperwork and supply tracking.
A Second New Tool Arrives the Same Day
ChatGPT Mil was not alone. The Department of War also launched Starshield AI's Grok for Government on GenAI.mil the same day, and per NOTUS, that tool brings deep reasoning inference, customizable workspaces, adaptive reasoning modes, and reusable playbooks built for acquisition market research and supply chain analysis. The idea, according to the Dallas Express report, is that having multiple AI models available reduces the department's dependence on a single vendor while expanding AI tools across the joint force.
The Chief Digital and Artificial Intelligence Office and the Office of the Under Secretary of War for Research and Engineering both say they will continue expanding the department's frontier AI capabilities, the outlet reports. That posture traces back to a decision the CDAO made in mid-2025, when it awarded prototype contracts worth up to $200 million each to OpenAI, Google, Anthropic, and xAI to build agentic workflows for national security missions, according to DefenseScoop — an $800 million bet against locking the military into any single company's technology.
From a Google-Only Pilot to a Multi-Vendor Platform
GenAI.mil did not start this way. The platform launched on December 9, 2025, powered solely by Google Cloud's Gemini for Government, and it attracted over 1 million unique users within its first month, per Defense News — a rollout Hoodline covered at launch. Adoption has only accelerated since: GenAI.mil has now onboarded more than 1.7 million unique users, according to the Dallas Express, and by February five of the six military branches had formally designated it their default generative AI platform for unclassified work, according to DefenseScoop, with the Coast Guard remaining the lone holdout because of its peacetime status under the Department of Homeland Security.
The Air Force went a step further, decommissioning its own homegrown chatbot, NIPRGPT, on December 31, 2025, and instructing 700,000 users to migrate their data and workflows onto the central platform, DefenseScoop reported at the time. Access to GenAI.mil remains tightly controlled: it is gated by Department of Defense Common Access Cards and limited to government-furnished equipment on the military's non-classified network, according to ClearanceJobs, which rules out public or unauthorized web access. The IL5 standard that ChatGPT Mil and Grok for Government now carry requires a FedRAMP High baseline, physical tenant separation, and U.S.-person staffing restrictions under DISA's Cloud Computing Security Requirements Guide, per Knox Systems.
Anthropic's Absence Points to a Legal Fight
Notably absent from Monday's expansion is Anthropic, whose Claude models were part of the same 2025 prototype awards but have not joined GenAI.mil. That gap traces back to a February dispute in which the Pentagon labeled Anthropic a national security supply-chain risk after the company refused to strip safety guardrails barring Claude from being used in autonomous weapons or domestic mass surveillance — a designation Hoodline covered as it unfolded in a fight with Defense Secretary Pete Hegseth.
That designation did not survive legal scrutiny. On August 27, U.S. District Judge Rita Lin in San Francisco ruled the Pentagon had acted illegally, finding the risk label amounted to unconstitutional retaliation, according to CBS News — a ruling Hoodline detailed in its report on the blacklist being struck down. While OpenAI and xAI adapted their models to meet Department of War requirements and secured IL5 accreditation, Anthropic's refusal to remove its safety limits set off the legal battle that culminated in that ruling just days before Monday's ChatGPT Mil and Grok launch.
For now, GenAI.mil operates strictly at the unclassified level, but the platform is widely seen as a testbed for a broader push to bring commercial generative AI onto classified military networks — a trajectory already visible in Google's separate negotiations, reported by Hoodline, to build a classified Gemini deployment for the Pentagon. The Department of War, for its part, says its overarching goal is straightforward: using artificial intelligence to support military missions and decision-making across the force.
