She posted videos of the menus on TikTok and ultimately chose fries and lemonade from vendors using conventional food photography. Onejeme described the abundance of synthetic menu graphics as an “AI flyer epidemic.”
“Not too long ago, people would just take a picture of the food and that would be on the menu. Now, these AI-generated pictures — the saturation, the coloring, the texture. It’s like you don’t even know what you’re going to get when it comes out,” she said. “At that point, is it false advertising?”
Similar examples have appeared internationally. AI-generated restaurant advertising spotted in Tokyo, Kuala Lumpur and Berlin has depicted food with unusual textures, exaggerated symmetry and shapes that bear only a loose resemblance to the dishes being advertised.
Julia Damphouse, a tour guide who photographed an AI-generated restaurant billboard in Berlin, said the image immediately appeared unnatural. “I see a lot of AI food and it just looks a bit shiny and strange, but there’s something particularly alien and wrong with this one,” Damphouse said. “There is something wrong with it. It looks too geometric. I imagine that’s it’s like if you take acid and food stops looking like food.”
Technology reporter Sam Biddle had a similar reaction after encountering an AI-generated shrimp scampi image at a deli. The dish contained circular, creature-like forms that looked nothing like conventional shrimp. “It wasn’t just unappetizing — which, you know, it is — but it was also sort of spiritually unsettling and cosmically unnerving,” Biddle said.
The reaction may have a psychological basis. Alexander Diel, a psychologist at Germany’s University of Duisburg-Essen, is part of a research team studying how people respond to synthetic food imagery. He said people are particularly sensitive to visual irregularities involving food because eating carries an inherent safety component. “Food is important to us,” Diel said. “At the same time, it’s something that can easily make us sick, so we are very sensitive to when something is even slightly off.”
A peer-reviewed 2025 study from Diel and other researchers found that slightly flawed AI food images were rated as more uncanny and less pleasant than either obviously unrealistic images or highly realistic food pictures. Another study published last month found that people expressed less desire to eat food shown in AI-generated images even when they rated its health and calorie characteristics similarly.
The researchers have proposed several possible explanations. Subtle visual abnormalities could activate instincts associated with avoiding unsafe food, while unfamiliar-looking dishes could also provoke food neophobia, or reluctance toward foods people do not recognize.
“We don’t know if the food is safe or not, so we’d rather be careful and not eat it,” Diel said.
The recognizable look of AI restaurant imagery may also reflect how image models produce and repeatedly reproduce familiar visual patterns. Reality Defender CTO Alex Lisle described the effect as convergence, in which generated outputs become increasingly similar as models encounter more synthetic material.
An experiment by an X user named Labtec illustrated that effect by repeatedly editing the same ChatGPT-generated restaurant menu 100 times. Across successive versions, the food became smoother, rounder and increasingly artificial. TechCrunch reported seeing similar deterioration when it repeated the experiment.
Lisle linked the recurring aesthetic to the kinds of restaurant images heavily represented in model training data, including menus from large fast-food and casual dining chains. When models repeatedly draw from similar patterns, the resulting independent restaurant menus can begin sharing the same polished visual style.
That consistency can become a disadvantage when customers interpret it as artificial. The problem is especially pronounced with food because consumers already know how familiar dishes are supposed to look.
The growing use of AI by restaurants helps explain why these images are appearing more often. According to the National Restaurant Association, 26% of restaurant operators are using AI, and marketing is the most common application among those businesses. The same industry is operating under financial pressure: 42% of operators reported that their restaurants were not profitable in 2025.
Earlier research suggested consumers were less likely to notice the difference. In 2023, restaurant ordering platform Slerp surveyed 100 people and found that more than 60% could not reliably distinguish AI-generated food pictures from real photographs.
At the time, Slerp founder JP Then argued that AI imagery could make sense for businesses with limited budgets as long as the images remained appetizing and accurately represented what customers would receive.
But greater exposure to generated imagery appears to be making synthetic food easier to recognize. Online users have increasingly circulated examples under descriptions such as “AI slop menus,” while food publications have criticized the images for making dishes appear unnatural or unappealing.
That backlash can have consequences for small businesses. One restaurant owner said an AI-generated menu image that spread online led to negative reviews and threats. The owner subsequently stopped using AI-generated food imagery. “It’s not fair to be grouped into this online hate. These people start giving fake reviews and trying to hurt these businesses because an owner used ChatGPT to make a poster,” the owner said.
Nancy Hopkins, president of the International Association of Culinary Professionals, said restaurants may find more useful applications for AI in operational tasks such as scheduling or payroll. Food photography presents a different problem because customers can directly compare the promotional image with the meal they ultimately receive. “I think it backfires if you’re a little startup restaurant. You might have all the good intentions in the world to use a photo and say we can serve you this, but really, when you get into it, you can’t,” Hopkins said. “No one can deliver what that photo brings.”
For restaurants unable to afford professional photography or design, Hopkins recommended simpler illustrations or smartphone photography rather than creating an idealized synthetic version of the dish. “Where food is concerned, people want to eat something that’s authentic and truly delicious,” she said. “Tell your honest story. Real people, real stories will always be better.”
This analysis is based on reporting from CNN.
Image courtesy of David Mack/CNN via CNN Newsource.
This article was generated with AI assistance and reviewed for accuracy and quality.
