
Google DeepMind chief executive Demis Hassabis, photographed in July 2025. Image: Christopher Michel (Cmichel67) / Wikimedia Commons, CC BY-SA 4.0, cropped
Google is losing researchers over its military AI work. Several Google DeepMind staff have resigned or are on their way out over the company’s contracts with the US and Israeli militaries, the BBC reported on Wednesday, and two of them have spoken on the record. One, London-based research engineer Sofia Liguori, quit this week and talked to the BBC even though her contract restricts her from doing so.
Liguori, who spent three years at the company, put it plainly:
Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google.
Sofia Liguori, former Google DeepMind research engineer, to the BBC
Who is leaving, and why
Liguori told the BBC she resigned over Google’s growing work with the US 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, left Google DeepMind earlier this month. 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”, or “to develop artificial superintelligence safely in the future”. His sharpest point was about control:
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.
Andreas Kirsch, former Google DeepMind researcher, to the BBC
Two current DeepMind employees, speaking anonymously, told the BBC that “a lot of people have left the company that feel strongly about this”. They are not the first to walk out of DeepMind on principle this autumn: in September engineer Robert O’Callahan quit because his team was making AI faster.
What Google signed
Google is one of several companies that signed deals with the US military this year to put their AI into classified settings. According to the BBC, Google’s agreement lets its models be used for “any lawful government purpose” and says they “should not be used for domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control”. OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and the start-up Reflection signed similar contracts.
The push comes from the top. Defence Secretary Pete Hegseth said in January that the US will become an “AI-first warfighting force”, and the Pentagon has since set up a command for autonomous warfare and gone shopping for AI agents that hack on their own. Anthropic’s Claude was dropped from Palantir’s Maven Smart System, the Pentagon’s main targeting platform, after Anthropic refused to lift its limits on military use.
Google’s AI Principles promise “appropriate human oversight” and alignment with “widely accepted principles of international law and human rights”. They used to go further. The 2018 version listed AI “we will not pursue”, including weapons and surveillance that breaks international norms; Google dropped that list when it rewrote the principles in February 2025, in a post co-written by DeepMind chief Demis Hassabis.
“You feel you are being watched”
Staff told the BBC they are afraid to speak out because their contracts forbid public criticism. “In principle I’m not allowed to criticise Google for the duration of my life,” Liguori said, adding that “talking about facts is not criticising”. Inside the company, she said, “people are scared of retaliation and putting their career at risk.”
Google defended its process. A spokesperson told the BBC: “We work with public sector organisations worldwide and follow a consistent process for reviewing and entering these contracts, including compliance with our acceptable use policies and AI principles.” It added that it has “strict workplace policies in place to prevent retaliation of any kind.”
An echo of Project Maven
Google has been here before. In 2018 thousands of its employees protested its work on the original Project Maven, a Pentagon effort to analyse drone footage with AI; some resigned, Google let the contract lapse, and it published those first AI principles. Its $1.2bn Project Nimbus cloud deal with Israel, shared with Amazon, drew staff protests in 2024. Jack Shanahan, the retired lieutenant general who led Project Maven, told the BBC the philosophy now seems to be to “go as fast as possible to deploy [AI technology] as widely as possible, potentially at the expense of the rigour and discipline of test and evaluation.”
Update: Liguori’s farewell letter
Liguori has since published her farewell letter, “Stay Human”, dated 20 October, her last day at Google DeepMind. Writing in a personal capacity, she says she had been on leave since Google signed its classified contract with the Pentagon in April, after a year of campaigning against Google’s work with Israel under Project Nimbus, with Lockheed Martin and with US immigration and border agencies.
Her main charge is about how such deals are written. She says they allow AI to be built into military and surveillance systems “while imposing no restrictions on permissible uses and leaving the company no monitoring power”, which she calls “a very convenient contract design for executives to escape accountability”. She also describes the mood inside: staff who raise military contracts with their teams, as she did, get “called into HR meetings on the pretext of having breached some internal policy”, which she calls intimidation.
Liguori does not urge others to follow her out. “I don’t think resigning is an effective method, certainly not as an individual act,” she writes, asking colleagues who stay to join the GDM Union, through which UK staff are organising with the UTAW and Unite unions. Its demands, she says, include “enforceable red lines on military/surveillance applications”, more transparency around contracts, the right to abstain on ethical grounds and a stronger sustainability commitment. Google hasn’t commented on the letter.
Update: Google’s new union deal leaves out DeepMind
Google UK has signed a union access agreement, with the Prospect union, on November 2, days after a new legal right for unions to reach staff at work came into force on October 30 under the Employment Rights Act 2025. Prospect can now run membership sessions at Google offices, including King’s Cross, and contact staff online. Google told Novara Media the deal covers its UK employees “but not including Google DeepMind”, and Prospect said DeepMind was left out because “other unions are making a recognition bid” there.
Unite and the CWU, whose joint UTAW branch backs the GDM Union, say they asked Google for access at midnight on October 30 and are still waiting for an answer. In a statement to members seen by Novara Media, they said Prospect’s deal offers only “occasional meetings and an internal page”, and accused it of breaking the TUC’s rules on poaching members. Prospect called the allegations “false”. Google said the agreement lets it meet the new rules “while preserving the respectful, direct dialogue our culture is built on”.
Update: Staff want their lifetime gagging clauses gone
Some London-based DeepMind staff have contracts that bar them from criticising Google for life, even after they leave, The Washington Post reported on November 3, based on a contract it reviewed and interviews with two current employees who asked not to be named. Liguori’s decision to speak to the BBC despite her own contract prompted UK colleagues to check theirs; some found the same wording, and dozens joined an internal discussion about it, one employee told the Post.
Google told the Post it had “updated our terms for new hires, which included removing the post-termination aspect of non-disparagement clauses”, and a Google executive told staff it would not enforce the clause against people who had left. The employees want it taken out of the contracts they have already signed, arguing that even a clause they think is unenforceable can scare colleagues into staying quiet. “The people who work on this technology have a role to play in shaping where it’s going,” one told the paper.
Update: The unions take their recognition bid to the state arbitrator
The unions behind the GDM Union campaign have turned to the UK’s legal route for recognition. Unite and the CWU applied to the Central Arbitration Committee for recognition at Google UK Limited on November 12, the committee’s case page shows; no decisions have been issued yet. If the committee accepts the case, it can order a ballot of the workers, and a majority of those voting is enough to win, with no minimum turnout, the campaign says.
DeepMind’s UK staff first asked Google to recognise their union voluntarily on May 4. Google declined but agreed to talks through the conciliation service Acas, according to the campaign, whose demands centre on AI safety, “unethical military contracts” and guarantees over AI-led automation. The filing comes ten days after Google’s access deal with Prospect, which left DeepMind out. The case page doesn’t describe the group of workers covered, and Google hasn’t commented. An earlier application by the same unions against Google UK was withdrawn in November 2025.
Why it matters
In 2018, staff pressure pulled Google out of military AI; this time the company is staying in and the dissenters are leaving. If the people most worried about how DeepMind’s models are used keep walking out, fewer voices inside the lab will be left to push back, just as those models move into classified war planning.
Sources: BBC News (interviews with Sofia Liguori and Andreas Kirsch, Google statement); Google AI Principles; Google, “Responsible AI: Our 2024 report and ongoing work” (February 2025); Google, “AI at Google: our principles” (2018); Sofia Liguori, “Stay Human” (20 October 2026); Prospect, “Prospect strikes landmark access agreement with Google” (2 November 2026) (Google statement); Novara Media (Unite and CWU statement, Prospect and Google responses); The Washington Post (via International Business Times; employee interviews, Google statement); Central Arbitration Committee, “CWU, Unite the Union & Google UK Limited (2)” (13 November 2026); UTAW, “GDM Recognition”.


