The Pentagon’s secure artificial intelligence platform now has ChatGPT.
The Department of Defense on Monday launched OpenAI’s ChatGPT Mil on GenAI.mil, making the chatbot available for work involving controlled unclassified information, or CUI — sensitive government information that is not classified but still requires specific safeguards.
The department said ChatGPT Mil is accredited for CUI at Impact Level 5, a DoD cloud-security standard for certain sensitive unclassified information.
ChatGPT Mil joins Google Gemini on GenAI.mil, giving military and civilian users access to multiple frontier AI models. A Defense Department official told Defense One that Gemini is useful for search, while ChatGPT can be used for more text-based work, allowing users to select a model for a particular task.
GenAI.mil is becoming a multi-model platform. The Defense Department also launched Starshield AI’s Grok for Government on GenAI.mil on Monday, saying the addition would help eliminate vendor lock-in and promote a broader American AI ecosystem.
GenAI.mil has already attracted more than 1.7 million unique users, according to the Defense Department. The platform launched nine months ago and is intended to serve the department’s more than 3 million personnel.
For the Navy and Marine Corps, GenAI.mil is already an enterprise requirement. A January Navy memorandum designated it as the Department of the Navy’s mandated CUI and IL5 generative-AI platform and directed Navy organizations and commands to transition to it by April 30.
Navy guidance states that GenAI.mil is authorized for CUI but prohibits the use of personally identifiable information and protected health information.
The initial ChatGPT Mil experience includes chat, files, projects and custom GPTs, with additional features expected over time, according to the Defense Department. The platform is designed for document-heavy work including planning, policy, logistics and administration.
An OpenAI official told Defense One potential military uses include disaster-response planning, resource allocation, reconciling supply and maintenance information and drafting transportation requirements.
The Defense Department has already cited a military example of GenAI.mil’s use. In May congressional testimony, the department said the Army’s XVIII Airborne Corps used the platform to produce a complete exercise operations order for the U.S. Southern Command area of responsibility in six weeks, with nine writers producing work that traditionally requires six to nine months by a much larger staff.
Preparing frontier models for GenAI.mil requires security testing involving the Defense Information Systems Agency and National Security Agency, according to a former defense official who spoke to Defense One.
GenAI.mil is an IL5 enterprise platform for unclassified work, not the classified AI capabilities the Defense Department is separately developing and testing for combat operations. Defense One reported that Pentagon and OpenAI officials do not intend to use models on GenAI.mil for combat; the department is developing AI capabilities for classified networks and warfighting missions.
OpenAI’s government sales chief Joe Larson said ChatGPT on GenAI.mil is intended to make unclassified work more efficient while maintaining safeguards, human judgment and accountability for high-stakes decisions.
OpenAI also said data processed on GenAI.mil remains isolated within the government environment and is not used to train or improve its public or commercial models.
Natalie Oliverio is an independent journalist and Navy veteran who covers military and veterans’ issues. Drawing on her experience as a talent executive and mentor, she writes about the policies, people and programs shaping the military community.
