Not long ago, personnel at the United States Department of Defense were granted access to the AI models ChatGPT and Grok. These are secure variants hosted on the Pentagon’s internal, centralized platform. With the integration of Gemini last year, the number of AI systems in use has now risen to three. What does this change? Why does the Pentagon need so many models?
Secure Versions Specifically Developed for the Department of Defense
Since the end of 2025, the United States Department of Defense has had its own Gemini AI version, named Gemini for Government. Developed by Google Cloud, this version represents the first AI tool integrated into the Pentagon’s centralized GenAI.mil platform. As stated in an official press release published on September 1, 2026, two new AIs are joining: ChatGPT Mil and Grok for Government. As their names indicate, these are secure versions of ChatGPT (OpenAI) and Grok (xAI). Naturally, these two versions have also been integrated into the GenAI.mil platform.
“I expect every member of the department to log in, learn how to use it, and immediately integrate it into their working methods. AI must be an integral part of your daily operational life; it must be your ally. By mastering this tool, we will gain the edge over our adversaries.”, said Secretary of War Pete Hegseth in a message to staff about GenAI.mil.
Why is the Pentagon Using Gemini, ChatGPT, and Grok?
It should be noted that GenAI.mil is a military infrastructure offering broad yet restricted access. Indeed, it is reserved for the three million military personnel, civil servants, and contractors of the United States Department of Defense. For now, the platform already counts around 1.7 million active users. It is worth recalling that the Pentagon prohibits its troops from using public AI tools due to confidentiality concerns. Thus, the special versions of Gemini, ChatGPT, and Grok discussed here have accreditation to handle sensitive government information without the risk of external leakage.
Moreover, it is important to emphasize that this innovation fits within a diversification strategy pursued by the Pentagon, aligned with a clear vision. Indeed, the aim is to transform the U.S. military into a highly efficient force, enhanced by AI. The central objective is to automate time-consuming tasks in order to allow troops to focus on mission-critical work. The AI models will thus be employed for administrative, logistical, and strategic missions. These tasks will include synthesizing policies and directives, managing the supply chain, planning transportation, and crisis management.
Finally, one might wonder why the Pentagon sought to diversify the AI models within its platform. In practice, the goal is to ensure always having the best tool for the right job. For example, ChatGPT excels at processing heavy documents, while Grok is particularly adept at market analysis and logical reasoning. For Defense, this diversification fosters innovation while safeguarding the military infrastructure from potential technical failures or service interruptions from a single provider.
