TLDR
Google DeepMind research engineer Sophia Laguardia resigned in protest over the Nimbus contract, a $1.2 billion deal providing Google and Amazon cloud and AI services — including the Gemini model — to the Israeli government and military with no usage restrictions. She describes discovering the contract after Google fired U.S. protesters in 2024, facing internal retaliation for raising concerns, and argues that current military integration of AI poses a greater immediate threat than hypothetical future rogue AI. Laguardia advocates for task-specific AI over general-purpose agentic systems and emphasizes tech worker organizing and unionization as essential counterpower to corporate decisions.
For you
PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.
Sign in to unlockSurprising Insights
Google was explicitly advised by both internal lawyers and an external consultancy not to include AI services in the Nimbus contract due to human rights risks, but signed it anyway with no usage restrictions.
The Nimbus contract legally prohibits Google from restricting how Israel uses the technology and obligates Google to secretly notify Israel of foreign court orders for data — a provision that actively shields military operations from legal scrutiny.
Laguardia argues that the dominant 'AI safety' narrative about superintelligent gods going rogue is convenient for leadership because it shifts accountability from human decisions about deployment to the technology itself.
She reveals that even 'soft' actions like taking a single day off to protest in front of the office would be a powerful signal if done by many workers, and that critical teams like Gemini or ChatGPT going on strike would force companies to listen.
Key takeaways
The Nimbus contract gives the Israeli military unrestricted access to Google's full AI suite, including Gemini, with no contractual limits on how the technology is used in warfare, and Google ignored internal legal advice to exclude AI services.
Google workers who raise concerns about military contracts face systematic retaliation: HR meetings, manager interventions, removal of internal posts, and use of vague policies to silence dissent, creating a culture of whispered solidarity but public fear.
Laguardia distinguishes her resignation from 'AI safety' concerns about future superintelligence, arguing that focusing on hypothetical rogue agents diverts attention from the concrete human choice to integrate unrestrained AI into military infrastructure today.
She advocates for task-specific AI applications — like her own work on optimal power flow — rather than energy-intensive general-purpose agentic AI, which she says is inherently less safe, more environmentally damaging, and designed to replace rather than augment human labor.
Tech worker organizing is expanding internationally across labs and companies, connecting anti-militarization, anti-surveillance, and labor rights struggles; Laguardia believes worker power — including the leverage of critical teams striking — is the primary check on corporate decisions, but automation threatens to erode that leverage.
The Nimbus contract includes a clause requiring Google to secretly notify Israel if a foreign court orders data handover, effectively shielding Israeli military data from international legal accountability.
Laguardia's decision was shaped by witnessing the Gaza genocide daily during maternity leave while knowing her company's technology was involved; she frames her resistance as consistent with everyday moral action, not exceptional activism.
Notable quotes
“I knew I couldn't stay there and continue with business as usual.”
▸ Notable quotes · 1:00“nothing about this contract was ever disclosed publi- publicly. And actually, all the public statement uh that Google uh did around this contract has been disproved as uh is was basically a lie.”
▸ Notable quotes · 4:03“they were advised not to include AI services and not to and to incorporate AI principles into the contract. What Google instead did is signing a contract which has no restrictions um about terms of usage and uh leave the company with very little visibility into how the products are used.”
▸ Notable quotes · 5:30“I no longer believe that our technology will have a net positive impact on society. I think there's a fantasy world where AI can help decarbonize grids, cut waste, protect ecosystems, and benefit humanity. But it's not this world and it's not this type of general purpose AI we are racing to develop.”
▸ Notable quotes · 9:34“when when the focus in on the model capabilities and and on the on the potential for agents going rogue, we are uh shifted shifting the attention from uh a a specific human choice, which is very irresponsible and needs to um carry responsibility uh and accountability, which is these unrestrained unregulated integration of these systems without um without red lines.”
▸ Notable quotes · 11:24“think about the pace at which these companies move. Um, think about Gemini team or ChatGPT team going on strike. They can't afford people going on strike.”
▸ Notable quotes · 16:55Chapters
0:45 — Resignation and Discovery of Nimbus: Laguardia explains she resigned after learning of the Nimbus contract — a $1.2B Google-Amazon deal providing cloud and AI services to the Israeli military — when Google fired U.S. protesters in 2024. She felt isolated in London, went on maternity leave, watched the genocide daily, and initially planned to quit without returning.
3:10 — The Nimbus Contract Details: Laguardia details the Nimbus contract: signed in 2021, provides full AI suite including Gemini to Israeli government and IDF with no usage restrictions, no public disclosure, and Google ignored legal advice to exclude AI services. Israeli officials have acknowledged Nimbus's importance for the Gaza war.
7:26 — Internal Organizing and Retaliation: Laguardia describes her role in sustainability projects, feeling she was 'ethics washing,' and facing systematic retaliation for raising concerns: HR meetings, manager interventions, post removals, and policy-based silencing. Colleagues expressed solidarity only privately due to fear.
9:33 — Resignation Letter and AI Safety vs. Current Use: Laguardia reads from her letter 'Stay Human,' rejecting the net-positive narrative of general-purpose AI. She distinguishes her protest — against current military integration — from hypothetical AI safety concerns, arguing the latter shifts accountability from human deployment choices to the model itself.
13:36 — Tech Worker Movement and Unionization: Laguardia discusses the rise of international tech worker organizing (No Tech for Apartheid, UTA union, Meta workers), connecting anti-militarization, anti-surveillance, and labor rights. She emphasizes worker leverage — especially critical teams striking — and the need for regulation to preserve worker power against automation.
17:54 — Vision for Ethical AI Development: Laguardia argues safe AI requires hard limits on deployment domains — no military or surveillance use — and advocates task-specific AI over energy-intensive general-purpose agentic systems. Task-specific AI augments human capabilities without replacement risks or loss of control.
20:30 — Personal Reflection and Moral Consistency: Laguardia reflects on her two-year-old daughter, framing her resistance as consistent with everyday moral action — not turning away from suffering — rather than exceptional activism. She hopes to explain the intensity of this period as staying true to her values in her workplace.
Speakers
Amy Goodman — Host
Sophia Laguardia — Interviewee