The New York Times is heading into a labor fight over how it uses AI on its own staff. According to The Verge AI, unionized workers with the Times Tech Guild say management refused to hand over information about how the company uses AI, what it plans next, and how those tools will reshape their jobs. The union filed an unfair labor practice charge earlier this month, and it’s not stopping there.
The Tech Guild, a NewsGuild of New York unit of roughly 700 engineers, designers, product managers, and data analysts, also filed grievances claiming the Times broke their contract by deploying two internal AI tools that track and grade employee performance.
This matters because the fight isn’t about AI writing articles. It’s about AI watching the people who build the paper.
The two tools at the center
The Verge AI reports the grievances target software called DX and Glean.
- DX bills itself as an engineering productivity tool. It tracks output, generative AI usage, and efficiency. The Times first pitched it internally as a way to measure the company as a whole, says Ben Harnett, a Times software engineer who chairs the unit’s generative AI committee. Over the past few months, the data got personal. Benchmarks are now applied to individuals.
- Glean pulls in wikis, GitHub docs, Google Docs, and emails so staff can search internal knowledge fast. The worry: a manager can query it about a single employee’s contributions, comments, and draft files.
Harnett’s account of how DX shows up in disciplinary meetings is the part that should make any knowledge worker pause.
“Now people in disciplinary situations are suddenly having read back to them, ‘You only did one pull request per week, per whatever, and that’s 25 percent below industry standard,'” he told The Verge.
He calls it a de facto quota. The metrics, he argues, flatten engineering work into an opaque score that says nothing about quality or how many features someone actually ships. The Tech Guild says the format of recent disciplinary notices suggests they were generated with Glean, a tool Harnett says also invents falsehoods and sends users on “wild goose chases.”
What the Times says
The company isn’t engaging on specifics. Spokesperson Danielle Rhoades Ha told The Verge the Times disagrees with the characterizations in the grievances and will respond through its “normal contractual process.” She noted the company has answered 80-plus requests for information from the Guild in recent years.
Worth flagging the irony: the Times uses AI for real journalism, parsing millions of Epstein-related documents and scanning satellite images of Gaza to locate a specific type of bomb. The dispute isn’t whether AI belongs in the newsroom. It’s who controls it and who gets watched by it.
Why this is bigger than one newsroom
This is the status quo cracking in real time. For years, newsroom AI debates centered on bylines, hallucinations, and disclosure. Now the bargaining table has moved to workplace surveillance, and the Times is far from alone.
- In April, 150 unionized ProPublica employees walked off the job for 24 hours, with AI disclosure a key sticking point.
- After McClatchy rolled out a generative AI tool that produces multiple versions of stories, some staff withheld their bylines in protest.
- The Times Guild, representing 1,500 editorial and support staff, is bargaining a new contract that demands a human behind any AI tool, clear labeling of AI-assisted journalism, and pay when the company strikes AI training deals.
What stands out here is the shift in framing. Tools sold as “developer experience” upgrades are landing as performance enforcement. Token counts and AI-usage frequency become pressure to do more, not better.
What to watch next
Harnett is clear that the union isn’t anti-AI. The position is that workers should have a say in how it’s deployed. “It’s going to distract you from actually doing a good job, which is what we think the company should want,” he said.
Expect productivity-tracking AI to become a standard line item in labor negotiations across tech and media, not a side issue. If you manage engineers or build these systems, the question coming your way is simple: does the metric measure work, or just police it? The Verge AI has the full report.