The US Department of Defense has launched its own versions of OpenAI’s ChatGPT and xAI’s Grok, giving three million civilian and military personnel access to generative AI tools specifically tailored for defence work. The two new tools, called ChatGPT Mil and Grok for Government, went live on Monday as additions to GenAI.mil, the Pentagon’s centralized AI portal that first launched last year with Google Gemini as its sole offering. The portal now provides three distinct frontier AI models under one roof, each adapted to handle sensitive government data without the collection practices common in consumer products. According to the department, GenAI.mil has already onboarded more than 1.7 million unique users, signalling strong internal demand for AI-powered productivity tools across the defence workforce. The deployment represents the most significant integration of commercial AI into the US military to date, touching everything from desk-based administrative work to frontline logistics and battlefield coordination.
What Each Tool Does
ChatGPT Mil, built through OpenAI’s OpenAI for Government programme, mirrors the interface that commercial ChatGPT users already know. The defence variant focuses on chat, file processing, project management, and custom GPTs, supporting document-heavy routine tasks such as administrative work, logistics, planning, and policy analysis. Additional features will be added incrementally as the tool matures. Grok for Government takes a more operationally ambitious approach. Delivered through SpaceX’s Starshield AI, a secure satellite communications network built on Starlink infrastructure, Grok is positioned as a tool that can help military personnel, acquisition professionals, and logisticians execute missions faster and with greater precision. The Pentagon described use cases ranging from market research analysis for procurement specialists to supply-chain management for frontline logisticians, and even broader applications in mission planning and intelligence synthesis. Both tools sit alongside Google Gemini, which remains available on the same portal. The three-model architecture gives the department flexibility to match tasks to the most suitable AI system, though the Pentagon has not disclosed how user traffic is distributed among them or whether certain roles are assigned specific models by default.
Security Without Compromise
The defining feature of all three tools on GenAI.mil is their exemption from the data collection practices that are standard in consumer AI products. In a commercial setting, user inputs are typically logged and sometimes used to improve future models. The military versions strip this out entirely, operating within a secure infrastructure designed to prevent sensitive defence information from flowing back to the AI providers. This security model addresses one of the biggest obstacles to AI adoption inside the Pentagon, where personnel previously had to choose between using consumer-grade tools that could expose classified or sensitive data, or avoiding AI assistance altogether. GenAI.mil eliminates that dilemma by providing frontier-level capability within a controlled environment. The portal’s rapid growth from zero to 1.7 million users in roughly a year suggests that the demand for AI productivity tools inside the defence establishment was substantial and largely unmet before the initiative launched. For many military analysts and planners, the ability to quickly summarize lengthy intelligence briefings or draft policy documents using a frontier model could save hours of manual work each week.
The Anthropic Question
The notable absence from the GenAI.mil roster is Anthropic’s Claude. The frontier AI lab was labelled a supply-chain risk by the Trump administration earlier this year after refusing to grant the Pentagon unrestricted access to its AI models, instead insisting on certain safety guardrails. Anthropic is currently fighting the designation in court, arguing that the restriction is politically motivated and harms national security by limiting the military’s access to top-tier AI. The dispute has effectively locked Claude out of the Pentagon’s ecosystem, leaving OpenAI, xAI, and Google as the three approved providers. For OpenAI, the ChatGPT Mil deployment represents a significant expansion of its government business and validates the OpenAI for Government programme as a viable channel for large-scale defence contracts. For xAI, the Grok deployment is arguably even more significant, linking its model to SpaceX’s Starshield network and embedding it directly into military logistics and operational workflows. The choice of Grok for satellite-linked operations also reflects Elon Musk’s deepening relationship with the US defence establishment, a connection that has drawn both praise and criticism from lawmakers and civil liberties groups.
A Broader AI Arms Race
The Pentagon’s AI push does not exist in isolation. Earlier this year, the department struck deals with Amazon Web Services, Microsoft, Nvidia, and Reflection AI to deploy AI on classified networks. The GenAI.mil expansion with ChatGPT Mil and Grok for Government represents the unclassified side of the same effort. Together, these initiatives are building an AI infrastructure that spans both the open and classified tiers of defence computing. The speed of adoption inside the department also carries geopolitical implications. As the US military integrates commercial AI tools into daily operations, allied nations are watching closely to determine whether similar models could work within their own defence establishments. The Pentagon’s experience with GenAI.mil is likely to become a reference point for NATO allies and other partners evaluating how to introduce frontier AI into military workflows without compromising national security. The launch of ChatGPT Mil and Grok for Government marks a turning point in the relationship between Silicon Valley and the US military, transforming the Pentagon from a cautious observer of consumer AI into an active user of its most capable tools. How the department balances the productivity gains against the risks of embedding commercial AI in defence operations will shape military AI policy for years to come.
discussion