Google DeepMind staff in London have voted to unionize in protest of the lab's AI being used by the U.S. and Israeli militaries, with 98% of Communication Workers Union members backing the move. The bid would cover at least 1,000 employees at DeepMind's London headquarters and, if recognized, would be the first union at a frontier AI lab. Workers have given Google management 10 working days to voluntarily recognize the CWU and Unite the Union before legal proceedings begin.
The vote follows Google's agreement last week letting the U.S. Department of Defense use Gemini inside classified military networks for "any lawful government purpose." More than 600 Google employees signed an open letter opposing the deal, which arrived alongside similar Pentagon contracts with OpenAI, xAI, Nvidia, Microsoft, and Amazon. Only Anthropic refused, and the Pentagon responded by labeling it a supply chain risk — a designation Anthropic is challenging in court.
The unionization push traces back to February 2025, when Alphabet removed a longstanding pledge from its ethics guidelines not to develop AI for weapons or surveillance. That pledge was originally added in 2018 after thousands of Google staff signed a petition over Project Maven, with several engineers resigning, eventually forcing Google to drop the contract. Workers want the commitment reinstated.
Key facts
- 0198% of CWU members at Google DeepMind voted to unionize, covering at least 1,000 staff at the London office.
- 02More than 600 Google employees signed an open letter opposing the company's deal letting the Pentagon use Gemini on classified networks.
- 03The unions gave Google management 10 working days to voluntarily recognize the CWU and Unite before legal proceedings begin.
- 04Workers want Google to reinstate a weapons-and-surveillance pledge that Alphabet quietly removed from its ethics guidelines in February 2025.
- 05Google fired more than 50 staffers in 2024 over protests of its military ties to the Israeli government.
"A lot of people here bought into the Google DeepMind tagline 'to build AI responsibly to benefit humanity,'" an anonymous DeepMind employee said. "The direction of travel is to further militarization of the AI models we're building here." Another staffer dismissed the language in the new Pentagon deal: "We think the [any-lawful-purpose] clause is vague enough to be effectively meaningless."
“Workers below VP level backed the union vote by 98%, and the bid would cover at least 1,000 staff tied to DeepMind's London office — the first union at a frontier AI lab.”— Jaeden Schafer
The letter to Google's UK and Ireland managing director Debbie Weinstein, lays out specific demands: an end to AI use by the U.S. Department of Defense and the Israeli military, an independent ethics oversight body, and the individual right for workers to refuse projects on moral grounds. Staff are also weighing in-person protests and "research strikes" — abstaining from improvements to core products including the Gemini assistant.
John Chadfield, the CWU's national officer for tech workers, framed the vote as recovering bargaining power that has eroded across the industry. "Fundamentally, the push for unionization is about holding Google to its own ethical standards on AI, how they monetize it, what the products do, and who they work with," he said. "Through the process of unionization, workers are collectively in a much stronger place to put [demands] to an increasingly deaf management."
DeepMind researchers told reporters the calculation has shifted since 2018. Cost-cutting, AI capex, and tech-sector layoffs have weakened the leverage individual employees once held. "One of the things we can look at through unionization is restoring that leverage," one researcher said. "If we can manage to get a seat at the table, whether that's in the ethics review, the AI review, deployments, or even on the Alphabet board, that's where we could restore leverage."
The Israel question runs through the campaign. In 2024, Google fired more than 50 staffers over protests of the company's military ties to the Israeli government, a memory that hangs over the current bid. "We don't want our AI models complicit in violations of international law, but they already are aiding Israel's genocide of Palestinians," one DeepMind employee said in a statement released by the CWU. "Even if our work is only used for administrative purposes, as leadership has repeatedly told us, it is still helping make genocide cheaper, faster, and more efficient."
Google has defended the contracts. Spokeswoman Jenn Crider said the company is "proud to be part of a broad consortium of leading AI labs and technology and cloud companies providing AI services and infrastructure in support of national security," adding that Google "remains committed to the private and public sector consensus that AI should not be used for domestic mass surveillance or autonomous weaponry without appropriate human oversight." The company did not respond to requests for comment on the union vote.
Recognition is not assured. The Alphabet Workers Union, formed by U.S. Google employees in 2021, is still not recognized by Alphabet for collective bargaining, though it has negotiated agreements for contractors. UK labor law gives the DeepMind workers a clearer statutory path — if Google declines voluntary recognition, an arbitration committee can compel it given the size and density of CWU support. The 98% margin and the 1,000-employee scope make a denial procedurally awkward.
Chadfield said other frontier-lab workers are watching. Both Anthropic and OpenAI announced large London expansions around the turn of the year, importing the same workforce now organizing at DeepMind. "The workers at other frontier labs have seen what Google DeepMind workers have done," he said. "They've come to us asking for help as well." This follows our coverage last week of the Pentagon's classified-network deals with Nvidia, Microsoft, and AWS, and Anthropic's exclusion from that round.
A recognized union inside a frontier AI lab would be the first real internal check on how these models get deployed in defense, and it lands at the exact moment the U.S. government is wiring them into classified systems. Google's leverage in that market depends on engineers willing to build for it; if a thousand of the most expensive ones in London can collectively refuse specific contracts, the political cost of every future Pentagon deal goes up. Anthropic's competitors just discovered that the supply-chain risk runs in both directions.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
