The cooking section of The New York Times has sparked an unexpected wave of subscriber revolt following the rollout of a new artificial intelligence feature. In an iOS-only beta test, the publication is now permitting subscribers to dynamically modify existing recipes using generative AI technology. Below the announcement article published on October 23, readers flooded the comments section with complaints, questioning the necessity of the tool, threatening cancellations, and demanding to know how to disable it. After four days and 172 overwhelmingly hostile comments, the moderation team ultimately disabled commenting on the piece altogether.
While subscriber metrics for the cooking division are not publicly disclosed, Jordan Cohen, the executive communications director for the Times, noted that the site and app recorded 514 million visits in 2025, alongside a YouTube following exceeding 1.4 million. Given a history of clumsy technological intrusions into culinary spaces, initial skepticism regarding the AI initiative was widespread. However, hands-on testing of the new functionality revealed that the system performs with surprising accuracy and practical utility.
Developing reliable, delicious recipes is an expensive and rigorous process that typically demands a dedicated editorial staff. Major culinary publishers like The New York Times Cooking and America's Test Kitchen rely heavily on meticulously tested recipes to maintain their brand integrity. Entrusting generative AI with a role in presenting this valuable archive of content carries substantial reputational risk. Earlier in the year, the publication utilized AI to assist with basic scaling and unit conversions, but the new customize this recipe field represents a much deeper integration.
When utilizing the feature, users can request modifications such as substituting ingredients or adapting dishes to specific dietary restrictions. For instance, requesting a banana bread adaptation without a standard loaf pan prompts the system to query alternative baking vessels and adjust the instructions accordingly. All modifications throughout the text are highlighted in a distinct red-orange typeface. If a user makes an unreasonable request, such as turning gochujang buttered noodles into a cookie, the system issues a hard redirect to a more suitable recipe.
Camilla Velasquez, the general manager of NYT Cooking, emphasizes that the tool is designed to act as an assistant rather than a primary creator, stating that users should not engage in open-ended chats with it. The underlying large language model processes the original recipe, the internal substitution guide, user commentary, and editorial expertise in a specific hierarchical order. Recipe editors established the foundational rule sets and subsequently evaluated the performance of the system to ensure it maintained the spirit of the original dishes.
Addressing concerns regarding the potential impact on the 60-member editorial staff, Velasquez stressed that the organization is actively expanding its team and maintains zero intention of reducing its recipe output. Current usage data indicates that subscribers primarily leverage the tool for ingredient and equipment substitutions, dietary adjustments, and scaling quantities to match available pantry items. Food writers and critics who approach the tool with cautious skepticism generally find that the measured implementation avoids many common pitfalls of culinary technology.
Testing the feature with Alison Roman's popular chicken and dumplings recipe demonstrated the system's nuanced capabilities. Requesting a substitution from bone-in chicken thighs to boneless breasts resulted in an appropriately retitled recipe complete with red-orange text indicators and a disclaimer noting it as an AI-modified version distinct from the creator's original work. The application successfully adjusted simmering durations and added butter to compensate for the lower fat content of breasts compared to skin-on thighs, though a suggestion to reintroduce shredded meat for an extended thirty minutes resulted in a less-than-ideal texture.
Subsequent tests involving a mushroom and broccoli grain bowl adapted to include squash and alternative cruciferous vegetables showed that the system successfully preserved the poetic voice of the original author, Francis Lam. Further experiments with yeasted waffles, pasta and chickpea stew, and sauerkraut preparation confirmed that the AI can effectively integrate adjustments such as extra vegetables, garlic, and proportional salt increases without disrupting the procedural integrity of the dishes.
Despite the initial resistance from the reader community, the feature provides genuine utility by helping home cooks utilize available ingredients and navigate dietary constraints. The Times has indicated that user feedback is being actively addressed, with plans to allow subscribers to toggle the AI feature off if desired. While the tool remains confined to an iOS-only beta environment, the strong bond between the cooking team and its readership suggests that human oversight and connection remain central to the platform's enduring appeal.
