“As of last week, we've had over 2 million people in a week across all three models log in and use AI,” Defense Department Chief Digital and Artificial Intelligence Officer Cameron Stanley told Punchbowl News, summarizing a surge in use of the Pentagon’s enterprise generative-AI service.
GenAI.mil adoption and user counts
GenAI.mil, launched in December, has moved quickly from a pilot to wide departmental use. Stanley reported more than 2 million people logging on and using the platform in a single week, a dramatic increase from the roughly 80,000 people across the department who were using AI in similar deployments before GenAI.mil rolled out. The platform’s first release week drew over 300,000 logins, and an earlier July milestone reached 1.7 million users.
The department employs roughly 3 million total personnel who can access GenAI.mil; the enrollment and activity numbers Stanley gave indicate a large fraction of that population has begun using the service for unclassified work.
The three hosted models: Gemini for Government, ChatGPT Mil, Grok for Government
GenAI.mil currently hosts three models: Google Cloud’s Gemini for Government, OpenAI’s ChatGPT Mil, and xAI’s Grok for Government. The platform provides vetted, internal AI tools intended for unclassified tasks such as analyzing information and creating and refining documents.
By naming the specific models available on the platform, the department has created a single access point to multiple vendor systems for approved civilians, contractors and military personnel.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildAgent Designer uptake and how agents are being used
One notable feature availability came in March, when Google Gemini for Government’s Agent Designer was added to GenAI.mil. Stanley said that within two weeks of that deployment users had created more than 50,000 agents. He emphasized these are not complex, robust agents but tools “helping people with their daily workflows,” automating rote tasks and streamlining routine work.
Across the platform, Stanley also said that by July users had created more than 100,000 custom agents, reflecting sustained uptake of lightweight automation built on the platform’s models.
Pentagon public posture on slowing AI development
Stanley framed rapid adoption as a success and pushed back on calls for limits. He characterized GenAI.mil as “a huge, a huge win for the department,” and said he initially believed AI adoption within the Pentagon “was going to be a harder sell.” On the question of slowing or pausing development—a debate that has drawn attention from lawmakers and tech executives—Stanley argued that “unilateral disarmament is the wrong approach.”
That stance sits alongside public remarks by President Donald Trump at the United Nations, where he said “we’re going to encourage it, not rein it in,” underscoring an executive branch message against pausing AI development.
How technologists, policymakers, and end users are likely to respond
- Technologists and security teams: The large, rapid user base and the creation of tens of thousands of agents will keep operational security and access controls central. The platform’s role as a vetted, internal repository for multiple vendor models ties vendors’ operational configurations to department security expectations.
- Policymakers and regulators: With senior department officials publicly resisting pauses in development, legislative and oversight actors concerned about the “dangers posed by advanced models” will have a clearer contrast to probe—deployment scale, governance controls for unclassified use, and the process for approving agents.
- End users and procurement leaders: The fast uptake of Agent Designer and high weekly logins signal demand for tools that automate routine workflows; procurement choices will likely emphasize vendor features that integrate into GenAI.mil’s vetted environment and ease user onboarding.
Stanley’s figures—2 million weekly users, a July peak of 1.7 million, more than 100,000 custom agents by July, and rapid Agent Designer uptake—paint a picture of institutionalizing generative AI across a large unclassified user base. The department’s choice to host multiple vendor models and to encourage internal agent creation has already reshaped who in the workforce uses AI and how. What remains to be seen, and is now a clear point for both internal managers and outside overseers, is how governance, security and oversight will keep pace with the platform’s growth and the broad push by senior leaders against pausing development.
