Google’s Pentagon Deal: AI for ‘Any Lawful Purpose’ and the Employee Backlash

Google has reportedly signed a classified agreement with the U.S. Department of Defense, allowing the military to use its AI models for “any lawful government purpose.” The Information broke the story, and it landed less than a day after Google employees publicly demanded CEO Sundar Pichai block the Pentagon from accessing the company’s AI, citing fears it would be used in “inhumane or extremely harmful ways.”

If confirmed, this puts Google in the same camp as OpenAI and xAI, both of which have already inked classified AI deals with the U.S. government. Anthropic was also on that list until it got blacklisted by the Pentagon for refusing to remove safeguards the DoD found inconvenient. The irony is thick: the company that once pledged “Don’t be evil” is now quietly enabling military AI, while employees are screaming from the rooftops.

The Employee Revolt

This isn’t Google’s first rodeo with internal dissent. Remember Project Maven? Back in 2018, thousands of employees signed a petition protesting Google’s involvement in a Pentagon drone program that used AI to analyze drone footage. That backlash was loud enough that Google eventually backed out and published a set of AI principles that explicitly banned using its technology for weapons or surveillance.

Fast forward to 2026, and those principles feel like a distant memory. The current deal is classified, so we don’t know the specifics, but “any lawful government purpose” is a wide enough net to catch just about anything the Pentagon wants. Employees are furious, and they’re not wrong to be. The same AI that powers Google Search or Gemini could theoretically be used to optimize drone strikes or analyze surveillance data. The company’s own AI principles say no to weapons, but a classified deal with the DoD is a loophole you could drive a tank through.

The Bigger Picture: Military AI Is a Gold Rush

Google isn’t alone here. OpenAI, which started as a non-profit with a mission to ensure AI benefits all of humanity, signed a deal with the Pentagon last year. xAI, Elon Musk’s venture, has also jumped in. The U.S. government is throwing money at AI for defense, and these companies are lining up to take it. The argument is always the same: “We need to stay ahead of China and Russia.” But the ethical lines get blurry fast.

What’s interesting is that Anthropic, which is often seen as the safety-first AI company, got blacklisted for refusing to remove safeguards. That tells you something about what the Pentagon wants: unfettered access to powerful models without restrictions. If Anthropic—a company built on safety research—balked, you have to wonder what Google and OpenAI agreed to.

My Take

I’ve been watching this space for years, and I’m not surprised. The money is too good, and the geopolitical pressure is too high. But Google’s situation is particularly hypocritical. They have a public-facing AI ethics board, they publish principles, they talk about responsible AI—and then they sign a classified deal that their own employees only find out about through leaks. That’s not transparency; that’s damage control.

The real question is whether these companies can maintain any credibility on AI safety while serving the military. Spoiler: they can’t. You can’t claim to build AI for good while selling the same tools to the Department of Defense for “any lawful purpose.” Lawful doesn’t mean ethical. Drone strikes are lawful. Cyber warfare is lawful. AI-enhanced interrogation is arguably lawful. The employees know this, and that’s why they’re pushing back.

Google’s response so far has been silence. Sundar Pichai hasn’t addressed the employee letter publicly, and the company isn’t commenting on the deal. That’s a bad look. If Google wants to keep its talent—and its reputation—it needs to have an honest conversation about where its AI ends up. Otherwise, “Don’t be evil” is just a slogan on a wall.