900 Google and OpenAI workers demand employers keep AI ethics limits on military use

Refused two specific Pentagon demands — mass civilian surveillance without judicial oversight, and autonomous lethal targeting without human authorization per decision — and said publicly: 'These are lines we won't cross regardless of the consequences.' After Trump declared Anthropic a national security supply chain risk, Amodei said 'disagreeing with the government is the most American thing I can think of.' His public framing gave the worker organizing movement its clearest moral anchor.

Announced OpenAI's Pentagon military AI contract on Feb. 27, 2026 — two days before Trump banned Anthropic. Publicly acknowledged the Anthropic ban set an 'extremely scary precedent' even while moving in the opposite direction. In an internal memo he said he personally shared Anthropic's red lines on surveillance and autonomous weapons — but signed a contract without them, a contradiction that workers cited as OpenAI's central credibility problem and that drove hundreds of OpenAI employees to sign the 'We Will Not Be Divided' letter.

Posted publicly on March 3 that mass surveillance violates the Fourth Amendment — an unusual constitutional assertion for a senior executive whose company was simultaneously in active negotiations for classified Pentagon AI contracts. His statement gave the internal employee letter immediate external amplification and signaled internal Google divisions about where ethical lines should fall, creating pressure on CEO Sundar Pichai that Google's silence on the Anthropic ban had not resolved.

Did not publicly respond to the employee letter or the Anthropic controversy. His silence stood in direct contrast to 2018, when he declined to renew Project Maven after employee pressure and published AI principles pledging Google would never develop weapons AI. By 2025 he had quietly dropped those pledges while pursuing Pentagon contracts — a reversal that the 'We Will Not Be Divided' letter explicitly named and called on him to reverse.

A key organizer of the 2018 Google walkout that killed Project Maven. By 2026 led the Signal Foundation and was a vocal public critic of the tech-military AI relationship. Her historical role gave the 2026 worker activism its institutional memory — she connected the current organizing to the 2018 precedent workers were invoking and explained publicly why Signal-based organizing was necessary after Google fired internal organizers in 2019 and 2023.

Provided the international advocacy context for the worker organizing, arguing that corporate ethical limits had become the only near-term barrier to fully autonomous lethal weapons because no binding international treaty existed. Her organization had pushed UN negotiations since 2013 without achieving a binding agreement — making the red lines that Anthropic drew, and that Google and OpenAI workers were demanding their employers maintain, the practical substitute for international humanitarian law.

Did not publicly respond to the employee letter or the Anthropic controversy. His silence contrasted with 2018, when he declined to renew Project Maven after employee pressure. By 2025 he had quietly dropped Google's published AI principles pledging it would never develop weapons AI — a reversal the March 3 letter explicitly named and demanded he reverse.

Did not publicly respond to employee letter or Anthropic controversy. His silence contrasted with 2018, when he declined to renew Project Maven after employee pressure. By 2025 he had quietly dropped Google published AI principles pledging it would never develop weapons AI, reversal that March 3 letter explicitly named and demanded he reverse.