Google’s military AI is now inside the Pentagon’s most classified networks. This did not happen overnight.
For years, Google kept the U.S. military at arm’s length because their employees protested. But in 2026, the company crossed a line it once refused to approach. Now, the rest of the tech industry is watching to see what happens next.
From “Don’t Be Evil” To Defense Contractor
In 2018, Google walked away from Project Maven, a Pentagon program that used AI to analyze drone footage. Over 4,000 employees signed a petition against it and about a dozen resigned.
But in early 2025, Google removed a key pledge from its AI principles. That pledge had committed the company to avoiding weapons technology. DeepMind CEO Demis Hassabis pointed to “a global competition for AI leadership” as justification. However, Human Rights Watch and Amnesty International both condemned the reversal.
What Google’s Military AI Actually Looks Like Now
In April 2026, Google signed a classified contract with the Pentagon. The deal lets the Defense Department run Google’s Gemini AI on classified military networks with the permitted use: “any lawful purpose.”
Other partnerships like OpenAI kept control over their safety settings. However, Google agreed to let the government adjust those settings directly. Critics called this a major concession.
Moreover, the contract did not appear from nowhere. By the end of 2025, Google had already deployed Gemini to 3 million Pentagon personnel through a platform called GenAI.mil. This new agreement simply goes further into classified territory.
Employees are Pushing Back
However, more than 600 Google and DeepMind employees signed an open letter. They urged CEO Sundar Pichai to reject the deal. Workers at Google DeepMind in the U.K. even voted to unionize in protest.
Still, the outcome looks different from 2018. Google now points to a straightforward counter-argument: every major competitor has already signed similar deals. Walking away now would mean surrendering and they’re unwilling.
The Industry is Falling in Line, Except for One
Furthermore, Google is not alone. OpenAI, Microsoft, Amazon, Nvidia, SpaceX, and Reflection have all signed agreements with the Pentagon.
Together, they will deploy AI across classified military networks. The Defense Department, recently renamed the Department of War, wants to build “an AI-first warfighting force.”
Only Anthropic has refused these terms. As a result, the Pentagon labeled Anthropic a “supply chain risk.” The military must stop using Anthropic’s products within six months but Anthropic is now challenging that decision in court.
The Governance Gap Behind Google’s Military AI Deal
Currently, Congress has not passed any laws governing military AI. So, companies set their own red lines with contracts that use language broad enough to allow wide flexibility.
In 2018, the fight was about one program. In 2026, it is much bigger and nobody outside the Pentagon can see how it ends.
