Google steps in after Anthropic tells the Pentagon no

Anthropic made a principled stand last week. The company told the Department of Defense that its AI models could not be used for domestic mass surveillance or autonomous weapons. That’s the kind of line you’d hope more AI companies would draw.

Google just drew a different line.

The search giant has signed a new contract with the Pentagon, effectively stepping into the gap Anthropic left behind. The terms aren’t fully public yet, but the implication is clear: Google is willing to go where Anthropic won’t.

This isn’t Google’s first rodeo with the military. Remember Project Maven? That was 2018, when Google supplied AI for drone footage analysis. Employees revolted, thousands signed petitions, and Google eventually backed away with a set of AI principles that supposedly banned weapons and surveillance.

Those principles look pretty flexible now.

Google’s current AI principles still say they won’t build weapons or surveillance systems that violate human rights. But the Pentagon’s interpretation of “human rights” and Google’s might not match up. The new contract doesn’t specify exactly how the AI will be used, but the timing—right after Anthropic’s refusal—speaks volumes.

I’ve been watching this space for years, and the pattern is consistent. When one company draws an ethical line, another rushes to erase it. Anthropic’s CEO Dario Amodei has been vocal about the risks of military AI. Google’s leadership has been more, let’s say, pragmatic.

The Pentagon isn’t going to stop wanting advanced AI just because one company says no. They’ll find another supplier. And Google, with its massive cloud infrastructure and DeepMind’s capabilities, is the obvious alternative.

What bothers me is the lack of transparency. Google’s new contract wasn’t announced with fanfare. It slipped out through procurement records and insider leaks. The company knows this is controversial. They’re betting that employees have shorter memories than they did in 2018.

Maybe they’re right. Google’s workforce has changed. Many of the activists who fought Project Maven have moved on. The new hires might not care as much, or they might not know the history.

But some of us remember. And the question that keeps coming back is: where is the line? If Google won’t hold it, who will?

Anthropic showed that a company can say no and survive. Their business hasn’t collapsed. Their reputation with privacy-conscious customers has actually strengthened. There’s a lesson there that Google seems determined to ignore.

For now, the Pentagon gets its AI. Google gets its contract. And the rest of us get to wonder what happens when the next ethical boundary gets pushed.