The New York Times NewsGuild is negotiating AI provisions in its contract, demanding human oversight for all AI-generated content, limits on AI-drafted stories, retraining programmes, and a share of licensing income from AI training data 1. Management has refused the licensing demand.
The licensing fight sits inside a larger contradiction. The Times sued OpenAI in December 2023 for using its articles to train language models without consent. The union is now asking the same employer that sued over unauthorised AI training to share revenue from authorised training — and management is saying no. The implication: The Times treats training-data income as a corporate asset, not a product of the journalists whose work the models consumed.
NYT tech workers, organised separately from the newsroom, struck for eight days and won a contract creating an AI impact committee — a formal seat when the company deploys tools affecting their roles 2. Labour Notes documented the same pattern across multiple early 2026 disputes 3: employers concede process (committees, consultation) before conceding substance (job guarantees, revenue-sharing).
With roughly 10% of US workers unionised, even favourable bargaining outcomes cover a narrow portion of the workforce facing AI-driven restructuring. For the remaining 90%, protection depends on legislation or on employers voluntarily extending the terms that unions extract — a bet with poor historical odds.
