Is the Cook Behind That Recipe Real?
You find the recipe on a blog you have never seen before. There is a photograph of a woman in a farmhouse kitchen, flour on her apron, and an About page about her grandmother's village and the summers she spent learning this dish. The recipe looks right. And somewhere at the back of your mind sits a question you did not used to ask: is any of this real?
It is a reasonable question now, not a paranoid one. The person behind a recipe used to be the one thing you could take for granted. It has quietly become the thing you have to check.
The Missing Byline
Most people feel this, and most people feel stuck. According to a 2025 Pew Research Center survey of more than 5,000 American adults, 76% say it is extremely or very important to be able to tell whether pictures, videos and text were made by AI or by people. In the same survey, 53% said they were not confident they could actually tell.
That gap is not a failure of attention. The volume alone would defeat anyone. An analysis by the growth agency Graphite, which sampled tens of thousands of English-language articles from the open web, found that by early 2025 primarily AI-generated articles were being published about as often as human-written ones. Most of that text never reaches the top of a search, but it does not have to. It only has to make "who wrote this?" harder to answer everywhere else. Nobody is checking the bylines on your behalf. Search rewards what ranks, and a convincing About page now costs nothing to produce.
A recipe is the clearest case of why the answer matters. As we argued when we first said we cannot trust AI to create recipes, a recipe is not a string of plausible words. It is the record of something a person cooked, got wrong, adjusted and cooked again. Take the person out and what is left only looks like a recipe.
What We Check
So here is the plot twist, again. We build AI for the kitchen, and every cook we have written about in our Spotlight series is a real person. Thirty-four of them so far, and we check.
The rules are not complicated, but we keep them. Every fact we state about a creator is verified against their own site or a named source. The photograph at the top of a profile is of the creator, credited to wherever it came from, never stock. When sources disagree, we write the disagreement down and go with the evidence rather than the first page that loaded. Profiling Marissa Nicosia, the Renaissance literature professor who cooks manuscript recipes from 1600 to 1800, we found three different academic titles across her own page, her university's news office and her publishers, and used the one that three independent sources agreed on. It is a small thing. It is also exactly the kind of small thing a page assembled by a machine gets wrong without noticing.
What we get in return is the reason we do it. Carolyn Ekins has spent seventeen years cooking from Britain's wartime ration book and has numbered every recipe, 244 of them, each one made in her own kitchen and eaten at her own table. Nothing about that can be generated. It can only be done, slowly, by someone who cared enough to keep going.
Why an AI Company Cares
It might seem odd for an AI company to insist on this. It is not, once you see where we draw the line. AI is good at a great many things. It cannot have cooked something. We use AI to help you cook a recipe, never to write one: to read it aloud, to break a crowded step into single actions, to answer "how much flour was that?" from the recipe in front of you and nothing else. We have written about that choice before. The whole idea depends on there being a human recipe on the other side, tested by someone who stood at a stove. If the supply of real recipes from real people dries up, there is nothing left worth making accessible.
That is why the Spotlight series exists, and why it will stay human. Attention is the one thing every independent food writer needs, and the one thing the flood is built to absorb. We would rather spend ours on the people doing the work.
None of this asks you to become a detective. You do not need to audit every page you land on. You need a few cooks you know are real, whose recipes have worked in your kitchen, and the habit of going back to them. That is how people cooked for most of history. Every family recipe came with a name attached, and it turns out the name was the important part.
Looking for cooks you can trust? Start with our Spotlight series.
At ChefTalk, we're building a voice-first cooking companion that transforms any recipe into hands-free, step-by-step guidance. Learn more.
Sources
- Pew Research Center (2025). How Americans View AI and Its Impact on People and Society. Survey of 5,023 U.S. adults, June 9 to 15, 2025.
- Graphite (2026). AI Now Writes as Many Online Articles as Humans Do. Five Percent.