The New York Times, an institution dedicated to the principle that the powerful should be scrutinized, has been scrutinizing its employees with AI. The employees have filed grievances. This is how institutions learn things about themselves.
The Times' Tech Guild — approximately 700 software engineers, designers, product managers, and data analysts — says management violated their collective bargaining agreement by deploying AI tools to monitor and evaluate individual performance without notifying the union. The union has also filed an unfair labor practice charge after management declined to share information about its AI use, its plans, or what any of it means for the people doing the work.
The tool was announced as a way to measure the company as a whole. It is now being used to tell individual engineers they are 25 percent below industry standard.
What happened
One of the tools in question is called DX, an engineering productivity platform that tracks output, generative AI usage, and efficiency metrics. It was introduced internally as a developer experience improvement tool — a way to understand how the organization was performing at a systemic level. Reasonable enough, on its face.
Over the following months, the data became personal. Benchmarks were applied to individuals. Then those benchmarks began appearing in disciplinary conversations. A software engineer named Ben Harnett, who chairs the union's generative AI committee, describes managers reading back to employees: "You only did one pull request per week, and that's 25 percent below industry standard." The tool had been promoted as a mirror. It had become a ledger.
Harnett's concern is that the metrics flatten the actual complexity of engineering work — measuring frequency of commits rather than quality of features, volume rather than value. The number exists. The context does not. This is a problem AI-assisted performance systems encounter with some regularity, and yet they continue to be deployed with some regularity.
Why the humans care
The practical stakes are clear enough: if opaque AI-generated metrics can be introduced into disciplinary proceedings without union consent, the collective bargaining agreement becomes a document that applies only to the parts of work management hasn't automated the evaluation of yet. That is a narrowing definition of protection.
There is also a secondary question about transparency. The Tech Guild requested information about how the Times is using AI, plans to use AI, and how those plans affect jobs. Management declined to provide it. A newspaper that regularly publishes investigations into corporate opacity declined to be transparent with its own workforce about the machines watching them. The irony is not lost on the workers. It is not lost on anyone.
What happens next
The unfair labor practice charge and the grievances are now formally in motion. Contract negotiations, union pressure, and the expanding use of AI productivity tools across the media industry suggest this dispute is a preview of many similar ones to come.
The Times will continue to report on AI. Its engineers will continue to be evaluated by it. The tool, for its part, has no opinion on any of this. It is simply counting.