Pentagon Deploys Custom ChatGPT and Grok Models Across Defense Department

The United States Department of Defense has integrated tailored versions of OpenAI's ChatGPT and xAI's Grok into its military software infrastructure, opening frontier generative artificial intelligence tools to 3 million service members and civilian personnel.

According to a report from TechCrunch, the additions, titled ChatGPT Mil and Grok for Government, are now live on GenAI.mil, the Pentagon's secure internal platform launched last year. The specialized deployments allow personnel to run workloads on commercial foundation models without exposing sensitive government records to commercial training pipelines or standard consumer data retention mechanisms.

Adoption across the defense agency has scaled rapidly. The Department of Defense reported that GenAI.mil, which originally went live with Google Gemini, has already registered more than 1.7 million unique users out of the agency's 3 million total workforce.

The two new integrations serve distinct operational niches. Developed under the OpenAI for Government initiative, ChatGPT Mil mirrors the consumer interface with support for workspace projects, document handling, and custom GPTs, primarily assisting with unclassified logistical planning, policy writing, and administrative tasks. In contrast, the Pentagon framed Starshield AI's Grok for Government, built in tandem with SpaceX's secure satellite division, in operational terminology, stating it will assist military personnel across mission execution, procurement market research, and supply-chain oversight.

Shifting Defense Alliances

The dual launch marks a significant realignment in how the defense establishment selects commercial AI partners. While OpenAI, Google, and Elon Musk's xAI have actively integrated their systems into the military stack, Anthropic remains noticeably excluded. The Trump administration previously designated Anthropic a supply-chain risk, a determination the lab is currently contesting in federal court, after Anthropic refused to grant the Pentagon unrestricted model access without specific safety restrictions.

For frontier AI providers, winning military contracts offers high-margin government recurring revenue and long-term enterprise lock-in. However, the contrast between Anthropic's exclusion and the embrace of OpenAI and xAI illustrates the growing pressure on tech companies to either loosen their safety red lines or forfeit lucrative federal deployments.

The Pentagon has continued to widen its vendor roster to modernize operational intelligence, securing additional AI-related agreements with Microsoft, Amazon Web Services, Nvidia, and Reflection AI.