A DeepMind researcher quit over Google's Pentagon AI deal — and he's not alone

Alex Turner spent more than two years working on artificial intelligence safety at Google DeepMind. In June, he walked away.

His reason: Google signed a deal with the U.S. Department of Defense — and that was a line he couldn’t cross.

Turner, a research scientist focused on AI safety, told Business Insider he resigned after learning Google had agreed to let the Pentagon use its AI technology for classified operations. The Defense Department confirmed in early May that it had reached agreements with Google, Microsoft, Amazon, and OpenAI for “legitimate military operational purposes.”

“I just thought: no,” Turner said in an interview.

Google’s partnership with the Pentagon has been simmering internally for months. In April, roughly 600 of Google’s 195,000 employees signed a petition urging the company not to enter any agreement involving classified work. At least one other employee posted a resignation letter internally in May, citing the same concerns.

The core tension is oversight. Classified agreements, by their nature, limit Google’s ability to monitor how its AI is actually being used — and some employees find that unacceptable. One DeepMind researcher posted on X that they felt “ashamed” of the Pentagon deal.

Turner had been uneasy since early this year. He began considering leaving in February, anticipating the Pentagon agreement was coming. Rather than just walk away, he tried to build guardrails.

Earlier this year, Turner proposed a framework for military AI applications that he hoped Google would adopt. His core demand was straightforward: AI-powered target identification and strike systems must always remain under human control. He submitted the proposal to DeepMind CEO Demis Hassabis, who directed him to two senior members of Google’s policy team for evaluation.

After a few rounds of discussion, the replies stopped coming. Soon after, the Defense Department confirmed the deal.

“That’s when I realized I couldn’t stay in good conscience,” Turner wrote in a blog post.

Turner had also caught the attention of Google’s top leadership. He had lunch with Google Chief Scientist Jeff Dean to discuss his military AI concerns. He helped organize an open letter to Dean, who had publicly supported Anthropic — a company that had itself clashed with the U.S. government over military AI use. The employee letter called on Google to establish clear red lines in its Pentagon partnership, including a prohibition on Gemini controlling autonomous weapons without human supervision.

The broader context matters. Earlier this year, Google updated its AI principles, removing language that had pledged not to develop AI for weapons or surveillance. Hassabis co-authored a blog post announcing the changes. When an employee asked at an all-hands meeting whether the principles had actually changed, Hassabis said they hadn’t. Turner found that hard to reconcile with the deleted language.

“If I can’t trust an easily verifiable claim, how can I trust the rigorous oversight he says is in place?” Turner wrote in an internal message to colleagues.

Turner is now doing independent AI safety research and looking for his next role. He’s been clear he won’t join another big AI lab.

“When an employee leaves a top AI lab, they usually go to another company,” he wrote. “They usually get a huge signing bonus. That’s not my choice. I didn’t reach out to competitor labs.”

He added: “I’m unemployed.”