AI Recipes Erode Public Trust in Online Cooking Content

The Guardian reports that Google's AI Mode search feature, which began rolling out in March 2025, has repeatedly generated recipes that merge steps and ingredients from multiple creators into inconsistent or unsafe "Frankenstein" instructions; in one case the model pulled content from a Reddit thread and recommended an inappropriate material. Fortune and Bloomberg interviewed 22 independent food creators who described traffic declines of 30% to 80% since Google's AI-generated summaries began appearing in search, with some, including Clean Eating Kitchen's Carrie Forrest, saying they have laid off staff as a result. Google told Fortune: "AI Overviews are often a helpful starting point to learn about a dish, but we see that people still want to go and read original recipes from creators. We're focused on making it easy for people to discover and visit useful sites that have a good user experience."

PPC Land and a side-by-side video test by food bloggers Adam and Joanne Gallagher found an AI Mode version of a key lime pie that used different ingredient quantities and methods than the bloggers' original recipe. CNN documented a home baker who followed a recipe later suspected to be AI-generated and spent hours on a failed bake. The observable failure pattern - unsafe mixing of sources, loss of fine-grained procedural constraints like timing and measurements, and surface-level photo-text correlations that look credible without being testable - maps directly onto known weaknesses of retrieval-augmented generation over noisy, heterogeneous training data.

Teams building instruction-generation or summarization for practical domains should treat this as an evaluation benchmark for actionable correctness rather than fluency alone. Concrete mitigations the reporting implicitly points toward include provenance-aware retrieval that keeps a single recipe's steps from a single source together, structured schema extraction for ingredients and steps rather than free-text merging, and targeted unit tests such as mass or volume conservation checks.

Search-ranking shifts for creator-owned recipe pages versus AI-aggregated results, the frequency of instruction-inconsistency errors as Google iterates on AI Mode, and whether provenance metadata becomes standard in AI-generated cooking content are the measurable signals to track, rather than assumptions about Google's intent.

Copyright law's weak protection for procedural instructions leaves recipe creators a narrow legal path, so the more durable fix is likely technical: attribution-preserving retrieval and stricter correctness checks in generation pipelines, plus platform-level traffic and revenue-sharing changes that Google has so far addressed only with directional statements about preserving referral traffic rather than concrete commitments.