Overview
- Google reached a classified agreement that lets the Defense Department run its Gemini AI on secure networks for “any lawful government purpose,” according to reports Tuesday.
- The contract language says the system should not be used for domestic mass surveillance or autonomous weapons without human oversight, yet it states Google cannot veto lawful operational decisions by the military.
- Pentagon AI chief Cameron Stanley confirmed expanded use of Google’s models for classified work and said the tools are saving thousands of work hours each week as the department diversifies suppliers.
- More than 600 Google employees signed an open letter urging CEO Sundar Pichai to reject classified workloads and to add stronger transparency and ethics oversight for military uses of AI.
- The agreement follows the Pentagon’s break with Anthropic over guardrails, ongoing litigation over that blacklist, and similar classified-use deals with OpenAI and xAI, while Google recently withdrew from a $100 million drone-swarm program after an internal ethics review.