AI-generated food images have become common in restaurant advertising, appearing on menus and sidewalk signs from delis to cafés. X user Maggie Moda's post spotlighting a series of oddly perfect AI food ads outside New York City eateries has racked up 2.8 million views, according to Fortune.
The appeal for restaurant owners is straightforward: professional-looking food advertising no longer requires professional food photography. For small businesses without the time or capital for a photo shoot, AI offers a cheap substitute.
When Grind & Unwind opened on San Francisco's Haight Street in May, owners AJ and Lyndsey Lozano put up a temporary menu with AI-generated food images while getting the new business running, SFGATE reported. A Reddit post mocking the images as 'Yum, slop' drew more than 670 upvotes. Lozano told SFGATE the images were meant to be temporary while the couple opened their first restaurant and cared for three children under 5.
Chi-Town Deli in Chicago offered a similar account. Manager Ali Malik told Business Insider the deli's AI-generated sandwich images were a placeholder while the restaurant waited for a new sign with actual photographs.
The legal question is not new, even if the technology is. In the 1965 Supreme Court case FTC v. Colgate-Palmolive Co., justices weighed advertisers using mashed potatoes as a stand-in for ice cream. Rebecca Tushnet, a Harvard Law School professor specializing in advertising and trademark law, told Fortune that advertisers can use unreal images as long as they don't mislead consumers about what they're buying. 'If a reasonable consumer would understand that the image was making actual claims about the composition or quality of the food, then that could be deceptive,' she said. If the image simply conveys the general kind of thing a customer will get, 'then it is not likely to be deceptive.'
Libby O'Neill, partner and deputy chair of advertising, marketing and promotions at Loeb & Loeb LLP, told Fortune that an image itself can make a claim about a product's size, quantity, ingredients, quality or appearance even without accompanying text. An 'AI-generated' label, she said, 'explains how the image was made; it does not correct a false impression about what is actually being sold.' Her standard: 'Sell the product you depict, and depict the product you sell.'
The numbers come first, and they tell a clarifying story. No new agency, no fresh rulemaking and no expanded bureaucracy were required to police this trend. A 60-year-old deception standard, built on the same logic that once governed mashed-potato ice cream, already covers photorealistic AI renderings.
What is actually disciplining restaurants is the market itself. Millions of viewers and hundreds of Reddit upvotes did more to correct AI 'slop' than any regulator could in the same window. That is the free-enterprise system working as designed: small businesses experiment with a cheap tool to cut costs, consumers push back when the tool misleads them, and owners like the Lozanos course-correct without a government mandate. The lesson for regulators eyeing AI advertising broadly should be restraint—existing false-advertising law, paired with an alert customer base, is already doing the job.
