Roughly 600 Google employees signed an internal petition urging CEO Sundar Pichai to walk away from a classified Pentagon AI contract, according to The Information. The Information reports the workers are pressing leadership to reject the deal outright, reviving a fight inside the company that hasn’t flared this publicly since the Project Maven backlash of 2018.
This is the first time in years that organized internal dissent at Google has targeted a specific defense contract at this scale. And it lands at a moment when every major lab is racing toward government and military revenue, not away from it.
What’s actually on the table
The contract in question is classified, so the public details are thin. What’s known:
- About 600 employees signed the petition asking Pichai to refuse the deal.
- The contract involves AI work tied to the Department of Defense.
- It’s a direct test of where Google’s red lines sit under its updated AI principles.
Earlier this year, Google quietly removed language from its AI principles that previously ruled out building AI for weapons or surveillance. That edit cleared the runway for exactly this kind of deal. The petition signers are essentially arguing the runway shouldn’t have been cleared.
Why this is bigger than one contract
To understand the stakes, rewind to 2018. Thousands of Google employees protested Project Maven, a Pentagon program using Google’s computer vision tools to analyze drone footage. The pushback worked. Google let the contract lapse and published its original AI principles banning weapons work.
The industry has moved hard in the opposite direction since then. OpenAI dropped its own ban on military use last year. Anthropic struck a deal with Palantir and AWS to bring Claude to defense and intelligence customers. Meta opened Llama to U.S. national security agencies. Microsoft has been embedded with the Pentagon for years through Azure.
Google’s leadership clearly decided sitting out wasn’t an option. The 600 signatures suggest a meaningful slice of the workforce disagrees.
What stands out here
Three things make this moment different from 2018:
- The competitive pressure is real. Defense AI is now a multibillion-dollar category. Refusing it means handing it to Microsoft, Palantir, and Anthropic.
- The labor leverage has shifted. After multiple rounds of layoffs across Big Tech, employee petitions don’t carry the same weight they did during the 2018 talent boom. Google can absorb the dissent in a way it couldn’t seven years ago.
- The principles already changed. In 2018 the protest forced a policy. In 2026 the policy was rewritten before the protest started.
That last point is the one practitioners should pay attention to. The fight inside Google isn’t just about one contract. It’s about whether internal ethics processes still constrain product decisions at frontier labs, or whether those guardrails have effectively been retired.
What to watch next
A few signals worth tracking:
- Pichai’s response. A flat rejection of the petition, even a polite one, sets the tone for how every other lab handles internal pushback on defense work.
- Whether senior researchers join. In 2018, departures from senior AI staff gave the protest real bite. If that pattern repeats, it changes the calculus.
- Contract specifics leaking. Classified deals rarely stay fully sealed. The use case will shape how the public reaction lands.
- Customer reactions. Enterprise buyers paying attention to AI ethics positioning may push back on whichever way Google moves.
The broader read: the era when frontier AI companies could credibly say “we don’t do defense” is over. What’s still being negotiated, contract by contract, is how loudly the people building these systems get to object when their employer says yes.
Full reporting on the petition and the contract details is available at the original source.