New York Times Employees File Unfair Labor Practice Charges Against Management Over AI Surveillance Tool
Should AI enter newsrooms—and if so, how? This controversy has now moved to the negotiating table. The New York Times NewsGuild, comprising approximately 700 engineers, designers, and data analysts, recently accused management of refusing to disclose plans regarding AI usage and its impacts, filing unfair labor practice charges in the process. The focus centers on two internal tools: DX, an engineering efficiency platform touted as "enhancing developer experience." Union president Ben Harnett stated that while DX was initially used for overall metrics, it has recently begun establishing individual benchmarks—employees have been cited during disciplinary meetings with statements like, "You completed only one pull request this week, which falls 25% below industry standards." The union contends this amounts to de facto quota setting, flattening the complexity of engineering work into a set of metrics that could easily be weaponized at any moment. The second tool, Glean, integrates company wikis, GitHub repositories, documentation, and email, enabling AI-powered queries. There is credible reason to believe that recent disciplinary notices were drafted using Glean's generated templates. Harnett noted that Glean can also produce fabricated content, potentially misleading users down irrelevant paths. The union asserts these tools fundamentally constitute surveillance technologies deployed against workers, violating multiple contractual provisions concerning privacy, monitoring, and collective bargaining. A spokesperson for The New York Times said they would respond through normal contract procedures. Currently, the guild is negotiating a new agreement, with core demands including mandatory human oversight for any AI tool, explicit labeling of AI-generated news content, and compensation for employees whose contributions inform AI model training. Harnett emphasized that the union does not oppose AI itself but opposes pressuring staff using metrics such as token consumption, stating, "This distracts you rather than helping you do your job well—which is precisely what companies should want."
