Pete Hegseth announced on Monday that the US Department of Defense is adding secure, military-grade versions of ChatGPT and Grok to its bespoke AI platform GenAI.mil. The move brings the two generative-AI tools to the department's 3 million employees, including both service members and civilian staff.
AI tools added to GenAI.mil
The platform, launched in December with a specialised version of Google's Gemini, now counts 1.7 million users and will soon host ChatGPT Mil and Grok. Both tools are configured to handle Controlled Unclassified Information at Impact Level 5, meaning they can process sensitive but unclassified data without feeding it back into the models.
"ChatGPT on GenAI.mil will make unclassified work throughout the Department more efficient, freeing up time and resources so DoD personnel can focus on the mission," said Joe Larson, vice president and head of government at OpenAI.
The Pentagon's press release added that ChatGPT Mil offers a familiar commercial experience within a secure environment, tailored to warfighter needs.
How the platforms will be used
Most day-to-day defence work is unclassified and involves document-heavy tasks in human resources, logistics and contracting. By leveraging ChatGPT's ability to summarise and analyse large texts, staff can accelerate routine processes. The military version of Grok, supplied by SpaceXAI's Starshield, promises "deep-thinking inference" and multiple reasoning modes, auto, fast and expert, to support activities ranging from market-research for acquisition officers to supply-chain management for logisticians.
Background and previous rollout
GenAI.mil's initial launch was criticised for its abrupt rollout. In December, a pop-up urging staff to "use AI" sparked concerns that the message might be a phishing attempt. A subsequent memo from Pete Hegseth clarified the initiative, stating, "The future of American warfare is here, and it's spelled AI," and urging all personnel to integrate the tools into their daily workflows.
Legal dispute with Anthropic
absent from the new lineup is Anthropic's Claude. Relations between the department and Anthropic soured after Pete Hegseth publicly criticised models that "won't allow you to fight wars," a reference to Anthropic's refusal to permit autonomous-weapon or surveillance applications. In February, a former administration ordered agencies to stop using Anthropic's technology, labeling the firm a "supply-chain risk". A federal judge later ruled that the department's designation was unlawful.
With contracts worth up to $200 million each awarded to OpenAI and SpaceXAI, the Department of Defense is signalling an "AI-first" mindset that aims to keep the United States at the forefront of military technology.