How the U.S. Military is Rapidly Deploying 100,000 Autonomous AI Agents: A Deep Dive

Translation result

[Digital Today reporter Chu Hyun-woo] Pentagon officials are rapidly deploying autonomous AI agents across the department’s unclassified network.

On May 4 (local time), TechRadar reported that the Defense Department used the agent-designer version of Google’s Gemini on its GenAI.mil platform to build more than 103,000 semi-autonomous agents in under five weeks. The department is now adding more than 20,000 new tools each week.

Those agents execute roughly 180,000 times per week — about 25,000 executions per day on average. A session counts as a single execution of an agent by a user.

The busiest agents carry out repetitive tasks such as drafting after-action reports, preparing official staff documents, analyzing images and reviewing financial and strategic materials. Field personnel are building tools directly on the network to automate routine digital work without requiring traditional programming skills.

Robert Malpass, the department’s deputy chief digital and AI officer for intelligence, told the INSA Spring Symposium that anyone across the department can build and use advanced AI tailored to their specific work context.

The system runs on the unclassified network under an Impact Level 5 operational approval and operates within defined security and oversight boundaries. Still, officials and experts caution about how quickly automation tools are spreading. Some warn that agents operating beyond tight Pentagon oversight could delete systems, disrupt services or act without clear human authorization.

Defense leaders say it is difficult to slow the pace of adoption. Andrew Mayes, the acting deputy chief digital and AI officer, said AI is accelerating technological change and that introducing new capabilities into the force should not take five to 10 years.

The case illustrates how quickly internal users are building AI tools themselves and putting them into service. It also signals a shift in how the defense sector adopts AI, with large-scale use occurring inside approved security boundaries.