Google DeepMind Employees Resign Over Military AI Contracts
Left 67%
Center 0%
Right 33%
2 left · 0 center · 1 right
What happened
Several employees of Google DeepMind, Google's London-headquartered AI laboratory, have resigned over the company's military contracts. London-based research engineer Sofia Liguori resigned this week after three years at the company, and AI researcher Andreas Kirsch resigned earlier this month; more than 600 Google workers had previously sent a letter to CEO Sundar Pichai urging him to block Pentagon use of company AI for classified operations. Earlier this year Google signed a Pentagon contract allowing its AI models to be used in classified military environments for "any lawful government purpose," while stating they "should not be used for domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight and control"; Google also holds a 2024 $1.2 billion Project Nimbus contract with Amazon providing Israel's government and military with AI technology and cloud computing. A Google spokesperson said the company follows a consistent process for reviewing public-sector contracts, including compliance with its acceptable-use policies and AI principles, and has dedicated channels for employees to raise concerns. OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and the startup Reflection signed comparable Pentagon agreements as U.S. Defense Secretary Pete Hegseth made integrating AI into the armed forces a priority, saying the Trump administration will become an "AI-first warfighting force."
Sourcing check · Fully on the record · 3 articles
3 of 3 articles cite a named source for the story. It was on the record from the first report.
What these terms mean and how they're counted
- On the record · names a source 3 of 3
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Sofia Liguori and Andreas Kirsch
- Primary evidence · cites a document or data 2 of 3
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: on-record letter, contract
- Credits another outlet 0 of 3
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 0 of 3
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Oct 21, 15:41 UTC by BBC, naming its source. First document within the hour, from The Intercept (on-record letter).
- Not counted 0
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out
Unpacked
Right-leaning coverage casts the resignations as a protest against Trump's Pentagon, while left-leaning coverage reports that a named engineer also cited Google's work with Israel's military. Liguori told the BBC she resigned over increasing work with both the US and Israeli militaries, and that account also describes a separate cloud deal that supplies Israel with AI and infrastructure. That Israeli military work, and the cloud deal, never appear in Breitbart's account, so a reader there meets only a fight over American defense contracts.
The moral language does not match. Left-leaning coverage, in the Intercept, has Liguori rejecting work that would "help kill people." Right-leaning coverage instead labels Google's staff an "extreme leftist employee population" uninterested in defending the country.
What classified tasks is Google's AI actually performing under the Pentagon contract?
Bottom line
The BBC has Sofia Liguori quitting over US and Israeli military AI, but Breitbart's version of her resignation never mentions Israel.
The receipts — 5 facts one side's coverage lacks
- Sofia Liguori said she resigned over Google's increasing work with the US and Israeli militaries
- Kirsch said he went to bed hopeful the employee letter would cause pause and woke to a worst-case version of the contract already signed
- Project Nimbus is a $1.2bn Google-Amazon contract providing Israel's government and military with AI and cloud infrastructure
- Google said Project Nimbus is not directed at highly sensitive, classified, or military workloads relevant to weapons or intelligence services
- Liguori said internal posts were deleted because they contained the word genocide, which she was told was an intrinsically violent word
The Left View
Left-leaning coverage presents the walkouts as a revolt against Google abandoning its own AI principles—human oversight and alignment with international law and human rights. Liguori told the BBC she feels "ashamed" and has "lost faith in my company being true to its stated ethical values"; she told The Intercept she could not work at a firm that would "help kill people" or make it "easier to escalate the brutality of war," and wrote internally that racing to deploy agentic AI "will definitely be a net evil." Kirsch said Google's "governance and oversight are not sufficient" for military and classified settings and that the company appears to have "no right to control or veto how the Pentagon uses its AI" and "limited visibility" into deployment. Coverage stresses targeting risk: a former OpenAI geopolitics lead, Sarah Shoker, said AI "compresses the time an operator has to make a decision" so that reviewing hundreds of AI-selected targets is "almost impossible," and retired Lt. Gen. Jack Shanahan, former director of Project Maven, described a Pentagon philosophy of going "as fast as possible" potentially "at the expense of the rigor and discipline of test and evaluation." The Intercept centers Israel and Gaza: Liguori described watching a "livestreamed genocide" and said internal posts using "genocide" were deleted because it was deemed "an intrinsically violent word"; she argued AI targeting lets systems define combatants, reduces people to "a set of aggregated data points," and creates an "accountability gap" in which everyone in the kill chain can say they did not decide. Employees told the BBC they fear speaking out—Liguori said "in principle I'm not allowed to criticise Google for the duration of my life"—and that concerns are routed into channels that return "semi-automated answers with absolutely no content."
The Right View
Breitbart recasts the same departures as ideological refusal to support U.S. defense. After summarizing the BBC account, it concludes that "Google's extreme leftist employee population is not interested in supporting America and the defense of our country," and directs readers to a MAGA-oriented book on keeping AI from "the leftists of Silicon Valley" and from China. It treats the Pentagon's push—and the fact that peer firms signed comparable deals—as ordinary national-security modernization, citing an under secretary's line that "speed defines victory in the AI era," and presents the earlier worker letter and Kirsch's public remark that the signed deal was a "worst-case version" of the contract and "shameful" as evidence of a workforce at odds with American military power rather than as a validation of the resignees' ethics claims.
Our Take (balanced)
The left's strongest argument is that Google cannot enforce its published principles once models enter classified military systems: Kirsch's claim of no veto and limited visibility, Shanahan's warning that the Pentagon is moving "as fast as possible" potentially at the expense of testing, and Shoker's point that AI-compressed timelines make meaningful human review of mass target lists "almost impossible," alongside Liguori's account of internal suppression of dissent. The right's strongest argument, as Breitbart frames it, is that the resignations show an "extreme leftist" workforce "not interested in supporting America and the defense of our country" while China competes for AI dominance, and that industry-wide Pentagon contracts are a necessary "AI-first" military project rather than an ethical breach. The central unresolved tension is whether classified military use of commercial AI can be reconciled with enforceable human control and the companies' own rules, or whether opposition to those contracts is a refusal to participate in national defense.
3 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.