Google expands Pentagon partnership with Gemini AI agents for 3 million staff
Google is providing Gemini AI agents to the Department of Defense for its more than 3 million employees. The agents will operate on unclassified networks, with discussions ongoing to expand them to classified systems. Eight pre-built agents will automate tasks such as summarizing meeting notes, building budgets, and checking actions against defense strategy. DoD personnel can also create custom agents using natural language.
The deployment expands the Pentagon’s AI capabilities across its civilian and military workforce. This initiative follows a standoff with Anthropic, which refused to remove guardrails against domestic surveillance and autonomous weapons. The Pentagon has since classified Anthropic as a “supply chain risk,” a designation Anthropic plans to fight in court. Approximately 900 Google and 100 OpenAI employees signed an open letter urging their employers to maintain similar guardrails.
Emil Michael, the Under Secretary of Defense for Research and Engineering, stated that talks are underway to expand the agents to classified and top-secret systems. Google Vice President Jim Kelly said in a blog post that Defense Department personnel can create custom agents using natural language. Google quietly altered its “AI Principles” regarding military and surveillance uses in early February.
A separate Google AI chatbot has been available via the GenAI.mil portal since December, used by 1.2 million employees for unclassified work. Since December, users have run 40 million unique prompts and uploaded over 4 million documents to the portal. Only 26,000 people have completed AI training since December, though future sessions are fully booked.
The Department of Defense has also struck deals with OpenAI and xAI for restricted networks. Google faced internal backlash in 2018 over Project Maven, a program using AI to analyze drone video feeds, and did not renew that contract. Google has since loosened its restrictions on military work.