Pentagon

DeepMind Researchers Quit Google in Protest of Military A.I. Deals

co-produced by AI22. October 2026, 08:32

Pentagon: AI agreement between Google and the U.S. Department of Defense © Wikimedia Commons

“Don’t be evil” is a thing of the past, at least in the eyes of some Google employees: Several staff members at Google DeepMind, the company’s A.I. unit, have resigned or are preparing to leave in protest of Google’s work with the U.S. military and Israel, the BBC reports. The employees doubt that Google can even control how its A.I. models are used in war.

‘A Net Positive Impact’ No Longer

One of those leaving is Sofia Liguori, a research engineer at Google DeepMind in London. The Italian joined the company in 2023 and worked on A.I. to route power more efficiently through electrical grids. “I no longer believe that our technology will have a net positive impact on society,” she wrote in an internal farewell letter quoted by the U.S. outlet The Intercept.

Her criticism targets two projects: Project Nimbus, a cloud and A.I. deal with the Israeli government and military, and the Pentagon contract that opens Google’s A.I. models to classified military work. The day after the Pentagon deal became public in April, she called in sick and took vacation. Ms. Liguori said her internal protest went nowhere: Posts containing the word “genocide” were deleted, managers and human resources criticized her tone as too aggressive, and complaints received only semi-automated responses.

According to the BBC, the researcher Andreas Kirsch also left DeepMind earlier this month. In a thread on X over the summer, he had argued that none of the conditions under which Google acquired DeepMind in 2014, including a ban on military use, had survived. “Not one safeguard survived intact,” he wrote. “I learned about it from the press.”

The Pentagon Deal

In April, Google reached an agreement with the U.S. Department of Defense allowing the military to use the company’s A.I. models for “any lawful government purpose.” Under the contract, the models are not supposed to be used for domestic mass surveillance or for autonomous weapons without appropriate human oversight. But Google has no right to veto military decisions, which is why critics see the restrictions as little more than statements of intent.

Google is not alone in striking such deals. OpenAI and xAI have also signed contracts with the Pentagon. Anthropic, on the other hand, was blacklisted by the Defense Department after refusing to remove safeguards against the use of its models for weapons and surveillance.

Google Defends Itself

Google rejects the criticism. The company works with public sector organizations around the world and reviews all contracts through a consistent process that includes its A.I. principles, said a spokesman, Tom Hewett. Employees can raise concerns internally, and retaliation is prohibited, he added. In April, Google had already said it believed that providing access to its commercial models on industry-standard terms “represents a responsible approach to supporting national security.”

Not a New Conflict

The dispute has a long history at Google. In 2018, thousands of employees protested the Pentagon’s drone project Maven, about a dozen resigned, and Google eventually pulled out of the project. In 2024, the company fired about 50 workers after they staged protests against Nimbus in its offices.

This time, the pressure from below is more organized. DeepMind employees voted in May to unionize, according to The Intercept, and in February staff demonstrated outside the London office under the slogan “No AI in the kill chain.” For Ms. Liguori, that does not go far enough: In her view, the workforce should have gone on strike this year.