Over 600 Google Employees Tell Sundar Pichai: No Classified Military AI

Another week, another internal revolt at Google over military contracts. This time it’s classified AI work for the Pentagon, and the pushback is coming from inside the building — specifically from DeepMind.

Over 600 Google employees signed a letter to CEO Sundar Pichai asking the company to block the use of its AI models for classified military purposes. The Washington Post broke the story, and the numbers are worth noting: the signers include more than 20 principals, directors, and VPs. That’s not just junior engineers firing off a petition. That’s senior leadership saying “this makes me uncomfortable.”

The letter itself is blunt. According to the Post, it states: “The only way to guarantee that Google does not become associated with such harms is to reject any classified workloads. Otherwise, such uses may occur without our knowledge or the power to stop them.” That last part is key — it’s not just about moral objections, it’s about losing control entirely. Once the work is classified, nobody inside Google gets to audit it, question it, or even know what it’s being used for.

This isn’t the first time Google employees have pushed back on military AI. Remember Project Maven back in 2018? That was the Pentagon drone imagery program that led to mass resignations and eventually forced Google to walk away from a lucrative contract. The company later published a set of AI Principles that explicitly banned weapons systems and technologies whose purpose is to cause harm. But principles are only as strong as the enforcement behind them.

The timing here is interesting too. Anthropic, one of Google’s key rivals in the AI space, is currently in a legal battle with the Pentagon over similar issues. That fight is still unfolding, but it shows the entire industry is wrestling with the same question: how close is too close to the military?

Google’s official position has been that it follows its AI Principles and reviews each contract case by case. But the employees’ letter suggests that process isn’t transparent enough. If the work is classified, who inside Google actually reviews it? And how do you ensure the principles aren’t being quietly bent in a windowless room somewhere?

I’ve seen this pattern before in tech companies. The leadership says “we have guardrails,” the employees say “we don’t trust the guardrails,” and the public never finds out who was right until years later when a whistleblower leaks something. The only real safeguard is public accountability, and classified work is the exact opposite of that.

Let’s be realistic: Google is not going to suddenly announce it’s cutting all ties with the Pentagon. The Department of Defense is a massive customer, and cloud contracts alone are worth billions. But the employee pressure does create a real tension. If enough senior people leave over this, it starts to hurt the bottom line in other ways — talent retention, recruiting, public trust.

What I find most telling is that the letter specifically calls out the “classified” nature of the work as the core problem. It’s not just about weapons. It’s about the inability to say no after the fact. Once you hand over a model and it gets wrapped in a security clearance, you’ve effectively signed a blank check.

Will Sundar Pichai respond publicly? The Verge’s story doesn’t mention a formal reply yet. But given the history of these internal letters at Google — from Project Maven to the walkouts over harassment settlements — silence usually doesn’t make the problem go away. It just makes the next letter louder.