The Department of Defense (DoD) is rolling out Google’s Gemini AI agents to its vast workforce of over 3 million civilian and military employees. This marks a significant expansion of AI use within the Pentagon, initially on unclassified networks. Discussions are already in progress to extend these capabilities to classified and top-secret systems, as confirmed by Emil Michael, the Under Secretary of Defense for Research and Engineering.
These AI agents aim to streamline various tasks. Eight pre-built agents will automate processes such as summarizing meeting notes, creating budgets, and ensuring proposed actions align with the national defense strategy. Jim Kelly, Google Vice President, stated in a recent blog post that DoD personnel will also have the flexibility to create their own custom agents using natural language prompts.
Google’s AI chatbot, accessible via the Pentagon’s GenAI.mil portal, has already seen widespread adoption. Since December, 1.2 million DoD employees have used it for unclassified work, generating 40 million unique prompts and uploading over 4 million documents. While adoption has been rapid, AI training has lagged, with only 26,000 individuals completing training. However, future training sessions are fully booked, indicating a growing interest and commitment from employees.
This expansion comes as the Pentagon aggressively seeks new AI partnerships. This push follows a disagreement with AI company Anthropic, which declined to remove safeguards against domestic surveillance and autonomous weapons from its technology. The Pentagon subsequently labeled Anthropic a “supply chain risk,” a designation Anthropic plans to challenge in court.
The incident sparked concern within Google and OpenAI, with roughly 900 Google and 100 OpenAI employees signing an open letter urging their companies to maintain similar guardrails. Google quietly updated its “AI Principles” in early February, addressing these specific uses.
In the wake of these developments, the DoD has also secured deals with OpenAI and xAI for use on restricted networks. Google itself faced internal opposition in 2018 over its involvement in Project Maven, a program utilizing AI to analyze drone video feeds. While Google did not renew that contract, it has since eased its restrictions on military-related work.
The Pentagon’s increasing reliance on AI reflects a broader strategy to leverage advanced technology for national security, despite past internal and external controversies surrounding AI ethics and military applications.
