The Pentagon’s centralized AI portal, GenAI.mil, now includes OpenAI’s ChatGPT alongside Google Gemini, following an official announcement today. The new model, designated ChatGPT Mil, is engineered specifically for use with controlled, unclassified information (CUI) and tailored to support defense personnel in tasks requiring text-based analysis and documentation. “Gemini excels for search and data scraping, while ChatGPT is better suited for complex text-based workflows,” explained a Defense Department official, speaking anonymously prior to the rollout. “Our goal is to provide options so users can select the tool that best fits their specific mission.”
<
Development of the military-grade ChatGPT began in February, with an initial deployment target set for early July. The DOD’s official release stated that ChatGPT Mil delivers a “commercial-grade experience within a secure DoD environment, optimized for warfighter requirements,” enabling document-heavy tasks in planning, policy development, logistics, and administrative functions. The platform is designed to scale and support over 3 million Department of Defense personnel, streamlining routine operations to free resources for critical mission priorities.
Potential applications include operational planning for Northern Command, disaster response coordination with agencies like FEMA, and logistical tasks such as supply reconciliation and transportation planning at combatant commands, OpenAI representatives noted. However, the model will not be used in combat scenarios, as OpenAI emphasized that its reliance on connectivity and compute-intensive operations limits its suitability for battlefield environments. The Pentagon continues to evaluate other tools for such specialized uses, avoiding conflicts over AI applications in operational contexts.
Security-first design
Like its Gemini counterpart on GenAI.mil, ChatGPT Mil incorporates layered security protocols to isolate CUI from commercial infrastructure. “We cannot operate CUI on public models,” stated a defense official, explaining that the system operates in a compartmentalized environment with distinct safeguards. Deploying the model requires extensive preparation, including securing computing resources, managing user identities, curating training data, and aligning with Department of Defense security standards. Former officials involved in testing highlighted the rigorous process, which includes dual reviews by the Defense Information Systems Agency (DISA) and the National Security Agency (NSA) to prevent vulnerabilities or unauthorized data exposure.
The GenAI.mil platform launched on December 9 with Gemini and later integrated agreements with additional vendors, including xAI. “ChatGPT on GenAI.mil enhances efficiency across the Department, allowing personnel to focus on mission-critical decisions through secure, accountable AI tools,” said Joe Larson, OpenAI’s government sales lead. This integration reflects a commitment to balancing innovation with stringent security protocols for national defense.
