New York Times Tests AI Search Summaries, Drawing Newsroom Pushback

The New York Times has quietly launched its first generative AI feature that reaches readers without direct journalist oversight—a search tool that produces AI-written summaries of Times reporting alongside article excerpts and links.

The experimental feature, currently visible to a limited group of visitors, represents a notable shift for a publication that has built its reputation on rigorous fact-checking and editorial standards. According to Times spokesperson Graham James, the tool aims to "create a better search experience" by helping users discover and engage with Times journalism more effectively.

Internal resistance from the newsroom

The rollout has triggered significant concern among Times journalists, particularly members of the editorial union. Jim Luttrell, Times Guild unit chair and senior staff editor, criticized the company's AI policies as "meaningless," arguing that management can "change and even violate the company's policy in any way they want."

The union learned of the AI summary testing during contract negotiations in July, according to reporting by the media newsletter Breaker. Union representatives have proposed that 22.5% of revenue from AI licensing deals be distributed among unionized employees, and they're pushing for mandatory human oversight of all AI applications—a requirement that would effectively block the current search implementation.

Luttrell pointed to past failures of AI summarization tools across the industry: "Every time AI summaries have been deployed, they have made embarrassing mistakes. That's why we need strong guardrails on AI in the contract—to protect readers' trust in our journalism."

Why it matters

The Times faces a unique challenge in adopting AI tools. While competitors and tech platforms have rapidly deployed similar features, the paper has positioned itself as a bulwark against AI-generated misinformation and a defender of verified truth. Introducing AI-generated text—even summaries of its own reporting—risks undermining that message at a moment when the distinction between human journalism and machine-generated content has become a core part of its value proposition. The internal tension also highlights broader questions facing legacy media: how to modernize digital products without compromising editorial standards that differentiate professional journalism from algorithmic content.

Broader AI experimentation

The search summaries aren't the Times' only AI experiment. Wirecutter, the paper's product review subsidiary, recently introduced Wirecutter Finder, which generates AI summaries and recommendations based on the site's existing reviews.

Several Times employees acknowledged to reporters that the paper's search functionality has long underperformed compared to its publishing tools, making improvements necessary. Yet the implementation raises practical concerns: the Times' political critics could exploit the AI tool by repeatedly attempting to generate inaccurate or embarrassing summaries, creating reputational risks for a publication that prizes accuracy above all.

These details were first reported by Semafor.