AI's first union arrives in London

LAST WEEK, 1,000 staff at Google DeepMind's London office voted 98% to unionize under the Communication Workers Union and Unite the Union, an outcome that, if recognized, would produce the world's first union at a frontier AI lab. The trigger was Google's agreement to let the Pentagon use its Gemini models on classified networks for "any lawful purpose," part of a broader pact in which six other leading AI firms — OpenAI, xAI, Nvidia, Microsoft, Amazon and SpaceX — signed similar terms. Anthropic alone refused; the Pentagon designated it a "supply chain risk" in response.

The workers' demands go past collective bargaining: reinstatement of Google's pledge — published after the Project Maven uproar in 2018, quietly removed from the company's website in February 2025 — not to develop AI for weapons or surveillance that violates international norms; an independent ethics oversight body; and an individual right to refuse contributions to projects on moral grounds. The pattern has clear antecedents. In 2018 four thousand employees signed a petition against Maven; Google declined to renew the contract. In 2020 and 2021 Timnit Gebru and Margaret Mitchell, the co-leads of Google's Ethical AI team, were dismissed. In 2024 Google fired 28 employees who protested Project Nimbus, the $1.2bn cloud-and-AI contract with the Israeli government. The workers' letter, addressed to Debbie Weinstein, Google's UK and Ireland managing director, gave management ten working days to recognize the union or accept mediated negotiations before a statutory recognition process begins; the CWU has separately threatened "research strikes" that would see engineers withhold contributions to Gemini and other core products.