DeepMind researchers quit over Google military AI contracts

Two researchers have resigned from Google DeepMind over the company’s military AI contracts, the BBC reported Wednesday. Both said Google’s oversight of how the Pentagon uses its AI falls short. Google said it reviews such contracts under a consistent process.

What happened

Sofia Liguori, a London-based research engineer, resigned from Google DeepMind this week after three years, the BBC reported. She spoke to the BBC in a personal capacity. The BBC said her contract restricts her from doing so.

“Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google,” Liguori told the BBC.

She said she left over Google’s growing work with the U.S. and Israeli militaries. “I don’t agree with this direction and this has led me to lose faith in my company being true to its stated ethical values,” she said.

Andreas Kirsch, an AI researcher, resigned from Google DeepMind earlier this month, the BBC reported. He told the BBC that “Google’s governance and oversight are not sufficient or meaningful enough to deploy AI safely and responsibly in military and classified settings.”

“Google seems to have no right to control or veto how the Pentagon uses its AI and also seems to have limited visibility into how it’s being deployed,” Kirsch said.

What the contract allows

Google signed a contract with the Pentagon earlier this year that allows its AI to be used for classified military work, according to the BBC. The agreement says Google’s models can be used for “any lawful government purpose.” It also says they “should not be used for domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control,” the BBC reported.

OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and the startup Reflection signed similar contracts, the BBC said.

What Google says

A Google spokesperson told the BBC the company works with public sector groups worldwide. “We … follow a consistent process for reviewing and entering these contracts, including compliance with our acceptable use policies and AI principles,” the spokesperson said.

The company also said it has “dedicated channels for employees to safely raise any concerns” and “strict workplace policies in place to prevent retaliation of any kind.”

Liguori told the BBC that at Google “people are scared of retaliation and putting their career at risk.”

Why it matters

The dispute turns on one question: who decides how a commercial AI model is used once a government buys it. The contract language the BBC quotes gives an answer on paper. Use is allowed for “any lawful government purpose,” with limits on domestic mass surveillance and on autonomous weapons without human oversight. Kirsch’s complaint is not that those limits are missing. It is that Google may have little power to see or enforce them after delivery. Google’s answer is that it reviews contracts up front against its own policies and principles. Both points can be true. A lawful-purpose clause leaves the judgment to the government customer, which is how most defense purchasing works. It also means a vendor’s ethics pledges rest largely on the buyer’s rules, not the vendor’s. Two resignations do not show a broad exodus. They do show that the people building these models want that tradeoff stated plainly.

What’s next

Similar deals across the industry mean the same oversight questions now apply to several AI companies. Kirsch told the BBC that current models “are not robust enough” for the battlefield.