By Hamza Faisal
Several researchers at Google DeepMind have resigned over the company's expanding military AI contracts, telling the BBC they no longer trust Google's safeguards for work that ends up in classified defence systems. The departures, revealed in a BBC investigation published on Wednesday, expose a widening rift inside one of the world's leading AI labs.
Their concern is practical: once Google's models are handed to the military, they argue, the company has little control over how they are used.
Two named exits — and warnings of more
Sofia Liguori, a London-based research engineer who spent three years at Google DeepMind, resigned this week. Speaking to the BBC despite contract terms restricting her from criticising the company, she said Google's growing work with the US and Israeli militaries left her ashamed to work there.
"Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google," Liguori said. She added that the direction had cost her faith in the company's commitment to its stated ethical values.
AI researcher Andreas Kirsch resigned earlier in October. He told the BBC that Google's governance and oversight were "not sufficient or meaningful enough" for classified settings, that current models are "not robust enough" for such uses, and that Google appears unable to track how its systems are used once delivered to the military.
They are not the first to walk out. In June, René Mayrhofer, then a director for Android platform security, resigned, writing that leaving had become "unavoidable" and that management had "lost its moral compass". Current staff told the BBC they fear retaliation for speaking openly, and others are said to be preparing to leave.
The contracts behind the backlash
At the centre of the dispute is an agreement Google signed with the US Department of Defense earlier this year permitting its AI models to be used for classified military work, for what the contract calls "any lawful government purpose". The deal reportedly bars autonomous weapons selecting their own targets and domestic mass surveillance without human oversight, but researchers say it still leaves the Pentagon wide latitude across logistics, planning and battlefield operations.
A second grievance is Project Nimbus, the $1.2 billion cloud and AI contract supplying Israel's government and military, signed in 2024. Google has said Nimbus does not cover sensitive, classified or weapons-related work, but staff anger over the deal has persisted alongside the Gaza war.
Google is far from alone: OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and Reflection have signed comparable deals, and US Defence Secretary Pete Hegseth has vowed to make the administration an "AI-first warfighting force". Critics question whether the technology is reliable enough for combat, particularly on military networks disconnected from the public internet, where safety measures and human supervision may not hold.
Google's defence — and what comes next
Google told the BBC it follows a consistent process for reviewing public-sector contracts, including compliance with its acceptable use policies and AI principles. "We also have dedicated channels for employees to safely raise any concerns — ranging from business ethics to misconduct — and have strict workplace policies in place to prevent retaliation of any kind," a spokesperson said.
The resignations leave Google with a credibility problem: the researchers building its most advanced systems now publicly question whether those systems belong in war. Anonymous staff told the BBC a pipeline of new hires could replace those leaving, potentially shifting the lab's culture. For now, the debate over military AI is no longer confined to outside critics — it is happening inside the companies building it.



