Google Gemini AI Agent Deployed to Pentagon Following ChatGPT Boycott Threats
Google has launched its Gemini AI agent for more than 3 million US Department of Defence civilian and military employees, according to Bloomberg. The AI agent will initially operate on unclassified networks, although discussions are underway to expand it to classified and highly secretive networks.
US Deputy Secretary of Defence for Research and Engineering Emil Michael stated that eight previously created agents will assist in automating tasks such as summarising meeting notes, drafting budgets, and verifying the alignment of proposed actions with national defence strategy. Google Vice President Jim Kelly revealed that Defence Department employees can also create custom agents using natural language, enabling AI customisation based on the specific needs of individual units.
Since December 2025, Google’s AI chatbot, available through the GenAI.mil portal, has been used by 1.2 million US Department of Defence (Pentagon) employees for non-classified work. To date, employees have executed 40 million unique prompts and uploaded more than 4 million documents.
However, AI training has not kept pace with this rate of adoption, with only 26,000 employees having completed training. Nevertheless, subsequent training sessions are fully booked, indicating growing interest in the technology.
This expansion comes amid Pentagon efforts to broaden AI partnerships, following Anthropic’s refusal to allow its Claude system to be used for mass surveillance of US citizens or in autonomous weapons systems. The Pentagon subsequently classified the AI company as a national security supply chain risk.
Around 900 Google employees and 100 OpenAI employees have signed an open letter urging their companies to maintain their ethical AI policies. Google quietly modified its AI principles regarding this specific use in early February.
The Defence Department has since made deals with OpenAI and xAI for limited networks. Google itself faced internal backlash over Pentagon work in 2018 when thousands of employees protested Project Maven, a programme that used AI to analyse drone video feeds.