Google quietly opens its AI to the Pentagon for ‘any lawful’ use

Google has reportedly signed a classified agreement with the US Department of Defense, allowing the military to use its AI models for “any lawful government purpose.” The Information broke the story, and it comes less than a day after Google employees publicly demanded CEO Sundar Pichai block the Pentagon from accessing the company’s AI, citing fears it could 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 US government. Anthropic was also on that list at one point, but got blacklisted by the Pentagon after refusing to remove certain safety restrictions the DoD found inconvenient.

Let’s be real here: “any lawful government purpose” is a phrase so broad it could cover anything from logistics optimization to drone targeting. And the fact that this deal is classified means we won’t know the specifics until someone leaks them or a FOIA request drags them out years later.

I’ve seen this pattern before. Google’s old “Don’t be evil” motto has been quietly retired for years, but this feels like another nail in the coffin. The employee backlash isn’t surprising either — Google’s workforce has a history of pushing back on military contracts, most notably the Project Maven controversy in 2018, which led to Google not renewing that particular contract. But times change, and apparently so did Google’s appetite for defense dollars.

What’s interesting here is the contrast with Anthropic. They got blacklisted for refusing to weaken their safety protocols. That’s a pretty clear signal that the Pentagon wanted access without the usual guardrails. Google, OpenAI, and xAI apparently had no such qualms, or at least negotiated terms that kept the deal alive.

Is this a problem? It depends on your perspective. If you believe AI safety is paramount and that military applications carry inherent risks of misuse, this is alarming. If you think the US military should have access to the best technology available for national security, this is just business as usual.

Personally, I’m uneasy. Classified deals with vague language and no public oversight rarely end well, especially when the technology in question is as powerful and unpredictable as modern AI models. The fact that Google employees are already protesting suggests the internal mood isn’t great either.

Either way, the genie is out of the bottle. The question now is whether Google — and its peers — can keep any meaningful control over how their AI is actually used once it’s in the Pentagon’s hands. History suggests the answer is no.