Google DeepMind Staff Resign Over Classified US Military AI Deals

Multiple researchers at Google DeepMind have resigned in recent weeks over contracts that allow the Pentagon to use the company's AI models in classified military environments. The departures highlight a growing gap between Google's stated AI principles and the practical control it retains once its technology enters government systems.

The contracts and the resignations

Earlier this year, Google signed agreements with the US Department of Defense permitting its AI models to be used for "any lawful government purpose" in classified settings. The contract language states the technology "should not be used for domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control" [1].

Sofia Liguori, a London-based research engineer who worked at DeepMind for three years, resigned this week. She told the BBC: "Realising my company is pushing more and more for AI systems in military operations makes me feel ashamed to work at Google" [1]. Andreas Kirsch, an AI researcher, resigned earlier in October. Alex Turner, a research scientist who worked on AI safety for more than two years, left in June after writing that senior management had insisted Google would not sign such a deal, yet the agreement went ahead [2][4].

In April, roughly 600 employees signed a petition urging leadership to avoid classified military work [2][3].

What the departing staff say

The core concern is practical control. Kirsch told the BBC: "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" [1]. He added that Google's governance and oversight are "not sufficient or meaningful enough to deploy AI safely and responsibly in military and classified settings" [1].

Liguori said she lost faith in the company being true to its stated ethical values, which include implementing AI with "human oversight" and aligning with "international law and human rights" [1][6]. Kirsch argues current models are not robust enough for battlefield use [1][3].

Sarah Shoker, formerly head of OpenAI's Geopolitics team, described a structural problem: "AI compresses the time an operator has to make a decision because it works much faster than a person, and that's especially risky in a high-stress environment," she said. "When you have hundreds of targets that are selected by AI and a short amount of time, it is almost impossible for a human operator to meaningfully review the AI's outputs for errors" [1].

Retired Lieutenant General Jack Shanahan, former director of Project Maven, said the "philosophy" seems to be: "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" [1].

Google's position

A Google 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." The company said it has "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" [1].

Google's published AI Principles emphasize "implementing appropriate human oversight, due diligence, and feedback mechanisms to align with user goals, social responsibility, and widely accepted principles of international law and human rights" [6].

Employee fears and the broader landscape

Liguori said her employment contract restricts her from publicly criticising the company "for the duration of my life" [1]. She and other staff told the BBC they fear retaliation and feel watched. A current DeepMind employee, speaking anonymously, told the BBC that "a lot of motivated employees who signed up for [the company's] original mission could leave – but there's also a pipeline of people at the start of the careers who could fill those positions. It could be a changing of the guard" [1].

The Google deals are part of a wider push. Similar contracts were signed this year by OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia, and the startup Reflection [1][7]. Defence Secretary Pete Hegseth has made integrating AI a priority, saying the US will become an "AI-first warfighting force" [1][7].

Anthropic previously held a classified contract but broke with the Pentagon after CEO Dario Amodei raised concerns about mass surveillance and autonomous weapons; the Defence Department then labeled Anthropic a "supply chain risk" [7].

Previous controversy

Google faced internal opposition in 2024 over Project Nimbus, a $1.2 billion joint contract with Amazon providing AI and cloud infrastructure to the Israeli government and military. At the time, Google said the project "is not directed at highly sensitive, classified, or military workloads relevant to weapons or intelligence services" [1]. A recent documentary, Naza, examined Israel's use of AI-powered targeting systems in Gaza; an Israeli minister called the claims false and distorted [1].

What remains unclear

The sources do not specify how many staff have resigned in total, nor what fraction of DeepMind's workforce they represent. The exact terms of the Pentagon contracts beyond the quoted passages are not public. Google's claim that its review process ensures compliance with AI principles cannot be independently verified from these sources. The practical effectiveness of the contractual restrictions on autonomous weapons and mass surveillance in classified environments is untested and, by definition, unobservable.