Throw a rock in New York City, and apparently you’ll hit an AI food ad.
At least that was X user Maggie Moda’s assessment after spotting a series of peculiar sidewalk signs outside the city’s delis and cafés in a post that has since racked up 2.8 million views. There were croissant sandwiches with almost comically perfect layers, an affogato with a side of random holes, and a chopped-cheese hero with cheese melted a little too perfectly over every piece of meat.
Generative AI is giving restaurants a cheap and remarkably easy alternative to photographing the food they sell. For small businesses without the time or money for professional food photography, that can be a real advantage. But the backlash to AI food ads reveals a paradox: Some customers would rather see an imperfect picture of a meal that actually existed than a polished one that didn’t.
That distinction matters beyond whether AI food looks gross. Restaurants have been making food look better in advertisements for decades, but generative AI makes it possible to advertise a photorealistic meal that never existed at all. And because that imaginary meal is being used to sell a real product to real customers, it raises an old question in a new way. How different can the food in the ad be from the food on the plate before it becomes deceptive?
