More than 600 Google workers write to Sundar Pichai to refuse Pentagon deal for classified AI work

Over 600 employees at Google have written to CEO Sundar Pichai refusing to support any Pentagon deal that involves classified artificial intelligence work, setting up a fresh internal flashpoint over the military use of AI.

The letter comes as defence agencies deepen their reliance on commercial AI systems, and as tech companies face increasing pressure, both from governments seeking advanced tools and from employees demanding ethical limits.

Employees write a letter to Google

In their letter, signed by over 600 workers including members of Google DeepMind, employees made it clear they do not want the company entering agreements that allow its AI to be used in classified environments, reports Washington Post.

Their concern centres on transparency. Classified work, they argue, would prevent the company from knowing how its technology is deployed, effectively removing oversight. That, in turn, raises the risk of involvement in activities they consider harmful, including autonomous weapons and mass surveillance.

More from Tech

“We want to see AI benefit humanity; not to see it being used in inhumane or extremely harmful ways. This includes lethal autonomous weapons and mass surveillance but extends beyond,” the workers wrote in the letter. “The only way to guarantee that Google does not become associated with such harms is to reject any classified workloads. Otherwise, such uses may occur without our knowledge or the power to stop them,” the letter said.

The employees said the only way to ensure Google is not linked to such outcomes is to reject classified workloads altogether. Without that boundary, they warned, the company could be complicit in uses it cannot see or control.

The protest also taps into a longer history of internal dissent. In 2018, similar employee pressure led Google to withdraw from a Pentagon drone imaging project and introduce principles limiting AI use in warfare and surveillance. Workers now fear those commitments are being eroded.

“Human lives are already being lost and civil liberties put at risk at home and abroad from misuses of the technology we’re playing a key role in building,” the Google employees wrote in their petition, without specifying which technology they were referring to.

Anthropic and Pentagon fallout

The backdrop to this resistance is the recent clash between Anthropic and the US Department of Defense, which has intensified scrutiny on how AI firms engage with the military.

Anthropic attempted to include contractual safeguards preventing its AI from being used in mass surveillance or lethal autonomous weapons. The Pentagon rejected those terms and subsequently barred the company from defence work, triggering a legal dispute.

The standoff has become emblematic of a broader struggle. Governments argue they need flexibility to deploy AI across a range of scenarios, while some companies are pushing for firm limits on how their systems are used.

For employees at other firms, including Google, the episode underscores the risks of entering defence partnerships without clear and enforceable boundaries.

But OpenAI was quick to shake hands with Pentagon

Unlike Anthropic, OpenAI moved quickly to secure a deal with the Pentagon for classified AI deployments.

CEO Sam Altman said the agreement includes safeguards, such as prohibiting domestic surveillance and ensuring human responsibility in decisions involving force. He also indicated that further contractual language would reinforce these protections.

Despite those assurances, the deal has sparked backlash. Reports point to a sharp rise in users uninstalling ChatGPT following the announcement, even as rivals like Anthropic saw increased traction.

The contrasting approaches highlight a growing divide in the AI industry, between companies willing to collaborate closely with defence agencies and those seeking to impose stricter ethical guardrails.

For Google, the pressure is now twofold. It must weigh the commercial and strategic benefits of defence contracts against the risk of internal unrest and reputational damage, as employees push back against any role in classified military AI work.