corporate VERIFIED
Mar 10, 2026, 00:00 UTC Pentagon

Google Deploys Gemini Agents on Pentagon Platform

Google deploys 8 Gemini AI agents on GenAI.mil for unclassified Pentagon work, marking a dramatic reversal from the company's 2018 Project Maven withdrawal over employee protests.

On March 10, 2026, Google deployed eight Gemini-based AI agents on the Pentagon’s GenAI.mil platform for unclassified defense work, marking the company’s most significant military AI engagement since its 2018 withdrawal from Project Maven. The deployment covered administrative automation, document analysis, logistics planning, and intelligence summarization across multiple Department of Defense offices. While the work was unclassified, the deployment represented Google’s formal return to Pentagon AI work after eight years of self-imposed restraint.

The 2018 Project Maven episode had been a defining moment for Google. After thousands of employees signed a petition and dozens resigned over the company’s involvement in a Pentagon drone imagery analysis program, Google pulled out of the project and published AI principles that explicitly committed the company to avoiding “technologies that cause or are likely to cause overall harm” and “weapons or other technologies whose principal purpose or implementation is to cause or directly facilitate injury to people.” The Gemini deployment did not technically violate these principles — administrative and analytical AI on unclassified networks fell comfortably within Google’s stated boundaries — but it signaled a fundamental shift in the company’s willingness to engage with defense work.

The timing was not coincidental. With Anthropic banned and its classified AI role being divided between OpenAI and xAI, the Pentagon’s unclassified AI portfolio was up for grabs. Google’s deployment on GenAI.mil positioned the company to capture a significant share of this work. Company leadership framed the engagement as “responsible defense AI” — a careful positioning that supported the Pentagon without directly crossing the red lines Anthropic had drawn.

The 300+ Google employees who had signed the solidarity letter supporting Anthropic two weeks earlier responded with a new internal petition opposing the GenAI.mil deployment. However, the petition garnered fewer than 50 signatures — a fraction of the 4,000+ who had protested Project Maven in 2018. The diminished opposition reflected both the changed composition of Google’s workforce and a broader shift in the technology industry’s relationship with defense work. The AI arms race had reshaped the political terrain on which these debates played out.

Sources