Google Signs Pentagon Gemini Deal Despite Employee Backlash
3 min readGoogle has officially signed an agreement to supply its Gemini AI models to the U.S. Department of Defense for use on classified networks, despite an open letter from more than 580 employees urging the company to refuse the deal. The contract covers Impact Level 6 and Impact Level 7 systems, allowing the Pentagon to deploy Gemini for “any lawful government purpose.”
The signing makes Google the latest in a wave of seven AI vendors brought into the Pentagon’s new GenAI.mil portal, an internal hub for accessing frontier models inside classified environments. It also lands at a politically charged moment for AI in defense, with the Pentagon publicly snubbing one major lab over its safety stance.
What’s in the Deal
According to SiliconANGLE, the Defense Department finalized procurement contracts on May 1 with seven companies: Google, Microsoft, OpenAI, Nvidia, Amazon Web Services, SpaceX, and Reflection AI. Notably absent from the list is Anthropic, which the Pentagon labeled a “supply chain risk” after refusing to remove guardrails that would have permitted the use of Claude in autonomous weapons and mass surveillance.
The Pentagon’s GenAI.mil portal will give cleared personnel access to commercial frontier models for analysis, code generation, and decision support inside air-gapped classified networks. Reflection AI, founded in 2024 by former Google DeepMind researchers, is the only newcomer on the list.
Inside Google, A Familiar Fight
Days before the signing, more than 580 Google employees, including senior DeepMind researchers and at least 20 directors and vice presidents, sent CEO Sundar Pichai an open letter asking him to bar classified Pentagon use of Gemini. As The Washington Post reported, the letter argued that on classified, air-gapped networks Google has no way to monitor how its models are deployed, leaving “trust us” as the only effective guardrail.
The contrast with 2018 is striking. Back then, roughly 4,000 employee signatures and a dozen resignations forced Google to walk away from Project Maven, the military’s drone-targeting program. This time, more than 580 names produced a public statement of company pride and a signed contract.
Why the Google Pentagon Gemini Deal Matters
The Pentagon’s procurement strategy now treats safety-first labs as outliers rather than partners. Anthropic’s exclusion sets a precedent that vendors who insist on restrictions around lethal use cases can be cut out of major federal pipelines, while rivals willing to operate under “any lawful purpose” terms are rewarded with access to classified networks and long-term contracts.
For Google, the bigger story may be cultural. The episode shows that the same workforce that once held veto power over military AI work has, in 2026, far less leverage. The next signal will come from how the company handles enforcement, audits, and use-case transparency once Gemini is deployed inside the Defense Department’s most sensitive systems.