Google DeepMind staff resign over military AI contracts

Staff resignations

The BBC has spoken to several tech workers who have resigned, or are leaving, Google DeepMind over its military AI contracts. Sofia Liguori, a London-based research engineer, resigned this week and spoke exclusively to the BBC in a personal capacity, despite her contract. "Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google," she said. After three years there, she said she had lost faith in Google being true to its stated ethical values, citing growing work with the US and Israeli militaries.

AI researcher Andreas Kirsch also resigned this month, saying Google's governance and oversight were not meaningful enough to deploy AI safely in military and classified settings, that it had no right to veto Pentagon use of its AI and limited visibility into deployment, and that current models were not robust enough, or trained, for such tasks.

Google's response

A Google spokesperson said the company works with public sector organisations worldwide and follows a consistent process for reviewing and entering these contracts, including compliance with its acceptable use policies and AI principles. It has dedicated channels for staff concerns and strict policies to prevent retaliation.

Contracts, Maven and oversight

Google is one of several tech companies that signed US military contracts this year to deploy AI in classified environments. Its Pentagon deal allows "any lawful government purpose", but says the models should not be used for domestic mass surveillance or autonomous weapons, including target selection, without appropriate human oversight and control. US Defence Secretary Pete Hegseth said in January the Trump administration would become an "AI-first warfighting force"; Emil Michael, an under secretary at the US Department of Defense, said: "Speed defines victory in the AI era."

Maven Smart Systems, the Pentagon's main platform for organising intelligence and data, was developed by Palantir and uses AI to help personnel select and identify targets to strike. Retired Lt Gen Jack Shanahan, former director of Project Maven, said the "philosophy" seems to be to deploy AI as fast and widely as possible, potentially at the expense of rigorous testing and evaluation. Sarah Shoker, formerly head of OpenAI's geopolitics team, said AI compresses an operator's decision time, and that with hundreds of targets selected in a short time, meaningful review for errors is almost impossible. Google's AI principles require human oversight and alignment with international law and human rights, but staff say oversight at critical moments is unclear.

Industry context

OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and startup Reflection signed similar contracts covering transport, planning and battlefield operations. In 2024 staff criticised Google over Project Nimbus, a $1.2bn joint contract with Amazon providing Israel's government and military with AI and cloud infrastructure; Google said it was not aimed at highly sensitive, classified or military workloads.

Employees told the BBC staff are scared to speak out because contracts prohibit public criticism. "In principle I'm not allowed to criticise Google for the duration of my life," Liguori said. A current DeepMind employee, who declined to be named for fear of retaliation, said many motivated employees could leave, but a pipeline of early-career people could fill the positions: "It could be a changing of the guard."

Sources: BBC