I Used NYT Cooking’s New AI-Powered Recipe Modifier, and It’s Actually Pretty Good
AI-assisted digest, checked against the original source before publishing. How we work
The New York Times Cooking app's iOS-only AI recipe modifier drew subscriber revolt — but the feature itself mostly delivers, testing shows.
The New York Times announced on October 23 that its Cooking app would use AI to let subscribers modify recipes, and the reaction in the comments section was immediate and hostile. Readers threatened to cancel subscriptions, demanded to know who had asked for the feature and how to disable it. After four days and 172 almost entirely angry comments, commenting on the story was shut off.
The backlash made WIRED's kitchen-technology writer expect to side with the detractors, having spent a decade watching smart-kitchen products fail: Bluetooth-connected appliances, apps built on mediocre recipes, toasters with optical sensors and touchscreens that still couldn't toast well. The suspicion is understandable because AI is often a poor chef, and tested, high-quality recipes are the core asset of both NYT Cooking and America's Test Kitchen.
The feature itself arrives as an iOS-only beta. A "customize this recipe" field sits above the ingredient list, and the Times has already used AI earlier this year for scaling servings and converting imperial measurements to metric. The new tool goes further: ask it to make banana bread without a loaf pan and it first asks what pans are available, then rewrites the method accordingly. Every alteration appears in visible red-orange text throughout the recipe.
Over several days of testing, swaps as simple as ground pork for ground beef and as complex as dietary-tailoring requests came out largely right. When pushed beyond the bounds of a dish — turning Eric Kim's gochujang buttered noodles into a cookie, for instance — the feature issues a "hard redirect" and points to a better-suited recipe, in that case Kim's gochujang caramel cookies. "We put a lot of effort into trying to tell you when not to do something if it's no longer in the spirit of the recipe," says Camilla Velasquez, general manager of NYT Cooking.
Scale explains the stakes. The Times does not disclose Cooking's subscriber count, but executive communications director Jordan Cohen said the site and app were visited 514 million times in 2025, and its YouTube channel has over 1.4 million subscribers. That is a large, vocal audience for a media brand whose identity rests on recipes its staff tests.
What to watch: whether the iOS beta expands to Android and the web, whether the red-orange change notation and hard-redirect guardrails hold up as requests get more ambitious, and whether a feature that works well in practice can still quiet subscribers who object to AI touching a tested recipe in the first place. The open question is less whether the technology performs than whether the audience will accept it.
Key facts
- • NYT Cooking AI recipe modifier announced October 23 as an iOS-only beta.
- • Commenting closed after four days and 172 almost exclusively angry comments.
- • NYT Cooking site and app visited 514 million times in 2025.
- • NYT's YouTube channel has over 1.4 million subscribers.
- • Times earlier added AI-powered recipe scaling and imperial-to-metric conversion.
Why it matters
Publishers are now exposing tested recipe libraries to generative modification, and the guardrail — refusing requests outside a dish's spirit — becomes the product's key design decision. Builders should note that technical accuracy alone didn't defuse the backlash; audience trust and visible disclosure of changes carried as much weight as output quality. Expect more beta rollouts with kill switche
Spotted an error? Send a correction.
Get the day's AI news in one email
The Daily Epoch · 60-second read · unsubscribe anytime