Over 600 Google workers urge Pichai to block Pentagon AI deal

The Pentagon wants Gemini behind closed doors, but Google's own engineers are trying to slam the brakes.

According to The Washington Post, over 600 Google employees signed an internal letter to CEO Sundar Pichai, demanding the company block the military from using its AI in classified settings. Organizers say many signers hail from Google DeepMind, with the roster featuring more than 20 principals, directors, and vice presidents.

Can't fix what you can't see

The protest follows a report from The Information revealing active talks between Google and the Pentagon over a classified Gemini deployment.

In the letter, employees argue that classified workloads eliminate all accountability. Once a model is tucked away in a classified enclave, staff won't know how the technology is deployed and will have zero power to prevent potential harms.

The pushback highlights a growing divide across the AI industry. Microsoft already provides AI tools in classified environments, and OpenAI renegotiated its own defense deal in February. Meanwhile, Anthropic is fighting the Pentagon in court after being designated a supply chain risk for refusing to weaken its safety guardrails, a stance that drew rare cross-industry support from peers at Google.

Why it matters

Classified systems operate in a total black box where public accountability vanishes and safety guardrails are nearly impossible to audit.

We think Google finds itself in a precarious bind. Walking away leaves lucrative defense contracts entirely to Microsoft and OpenAI. But moving forward risks alienating DeepMind researchers who signed up to build frontier intelligence, not secret military tools.

With Anthropic showing that a lab can legally challenge the Pentagon rather than yield, Pichai's next move will test whether Google’s stated ethical boundaries actually hold when real government dollars are on the table.