The Pentagon now has its own version of ChatGPT and Grok, serving 3 million personnel

The popularity of the GenAI.mil portal is striking. According to data published by the Department of Defense, the portal has already onboarded more than 1.7 million unique users out of the department's 3 million personnel — meaning more than half of the department's workforce is already using the generative AI platform. These latest additions are part of the Pentagon's broader effort to use AI tools to accelerate work and give the Department of Defense an edge without sacrificing security. ChatGPT Mil comes out of the company's 'OpenAI for Government' program and will offer an experience familiar to commercial ChatGPT users, supporting chat, files, projects, and custom GPTs. According to the Defense Department, the tool will focus on document-heavy, routine unclassified work such as administrative tasks, logistics, planning, and policy, with other features added over time.

Compared with ChatGPT Mil's 'administrative and logistics' positioning, the Pentagon describes Grok's capabilities in much broader and more militaristic terms. The report quotes that 'Starshield AI's Grok for Government will provide the warfighter...' — Starshield is SpaceX's satellite business, meaning the integration of xAI and SpaceX has extended into the defense sector. The two companies are forming a combination of 'AI models + satellite communications' in the government market. At the same time, the news highlights Anthropic's absence: its Claude model does not appear in the GenAI.mil portal. The reason is that Anthropic refused to give the Pentagon unrestricted use of its AI tools and instead insisted on certain safety guardrails, after which the Trump administration labeled it a supply-chain risk — a designation it is currently fighting in court. Anthropic has become the only major U.S. AI lab with an open confrontation with the government over defense procurement.

This event offers several lessons for the AI industry and broader technology policy. First, 'government customization of commercial AI' is becoming a new competitive arena for major players: OpenAI, Google, and xAI are entering the defense market through secure portals, showing that government customers have shifted from 'bystanders' to 'large-scale buyers.' Second, the conflict between AI safety principles and commercial interests is now out in the open: Anthropic's insistence on safety guardrails collides head-on with the government's demand for unrestricted use, and the outcome of this lawsuit will define the negotiating boundary between safety and business for AI companies. Third, data sovereignty is being redefined in the AI era: military versions of AI tools are exempt from data collection, meaning 'whose data flows where' is becoming a core consideration in government AI adoption. For ordinary observers, the Pentagon's full embrace of generative AI marks the extension of AI technology from consumer and enterprise applications into the most central domain of national security.

For readers following AI technology adoption, the Pentagon launching custom versions of ChatGPT and Grok is one of the most iconic 'AI + government' events of late August 2026. It reveals a clear direction: once commercial AI models reach a certain level of safety and compliance capability, governments — especially defense departments — become some of the largest buyers. OpenAI secured its entry ticket to the defense market through the 'OpenAI for Government' program, xAI is positioning itself across 'models + satellites' through SpaceX's Starshield, and Google's Gemini was already there first. Anthropic's choice, meanwhile, reminds the industry that while pursuing commercial growth, the insistence on safety principles still has value — even if the cost is losing government orders in the short term. What's worth watching next: where the lawsuit over AI safety and defense procurement goes, how the usage boundaries of military AI tools are drawn, and whether the promise of 'data collection exemption' holds up in long-term operation. Either way, the relationship between generative AI and national security has moved from discussion to practice.