AI Food Images Have A Business Upside. They Also Have A Trust Problem
Benjamin Claeys is CEO of QR TIGER, MENU TIGER and GiftLips. He also hosts Stay QRious, a podcast about QR code best practices.
gettyA few weeks ago, I walked into a canteen and ordered tacos. The photo showed a full plate loaded with fresh toppings. What arrived was different: smaller portions, fewer toppings and a flat presentation.
It wasn’t a terrible meal. But the photograph had promised something else.
That mismatch is becoming more common as restaurants turn to AI-generated food images. Operators can now generate food photos quickly, sometimes without the high cost or lead time of professional food photography.
The economic pressure is real. U.S. restaurant sales are projected to reach $1.55 trillion in 2026, even as operators continue to face uneven traffic and rising food and labor costs.
When every dollar counts, faster and cheaper visual content is understandably attractive. But a menu image isn’t just decoration. It’s a direct promise to the customer about what they’ll receive. When that promise is broken, the short-term savings come at the cost of customer trust.
There’s another problem with synthetic food imagery: It doesn’t make the meal more appealing.
A 2025 study in Appetite examined how people responded to AI-generated food images and found evidence of an “uncanny valley effect.” Images feel unsettling when they look highly realistic but don’t accurately reflect the food in reality. When an AI burger looks too smooth or sauce looks like plastic, the human brain notices. Instead of making people hungry, it makes the food feel less appealing.
A CNN reporter noted a similar reaction from an ad sales worker who, while browsing a Hollywood market, described AI vendor images as having a “too-perfect, almost rubbery aesthetic” that made her doubt the quality of the actual food.
Traditional food photography has always involved impeccable lighting and styling to make real food look good. The difference with generative AI is that it invents items from scratch.
Generative AI tools learn from large amounts of text and visual data to create realistic-looking images, but doesn’t maintain the physical consistency of the food they depict. An AI generator can easily add bigger portions, extra toppings or textures.
This creates legal risk. The Federal Trade Commission (FTC) evaluates advertising based on the overall impression an image gives to customers. If a digital menu photo creates an expectation that the kitchen can’t meet, it contributes to a misleading advertising claim.
After years of studying the restaurant industry, I’ve seen how quick they adapt: They add new dishes, update seasonal menus and move their menus online, often without the time or resources to create everything from scratch.
AI can help remove some of that friction. AI-assisted menu digitization helps restaurants convert existing printed menus into digital versions faster. AI interprets the restaurant’s existing dishes, prices, descriptions and structure, then provides images while owners review and edit the results. It speeds up conversion without inventing menu content.
Cost is another factor. For independent restaurants and startups, professional food photography can be expensive. A small food photography shoot is estimated at around $350 to $500, while AI-generated images provide a lower-cost temporary option. Restaurants can also use AI to test different plating, backgrounds or visual concepts before investing in a photoshoot.
The key is knowing where AI fits. Once real photos are available, replace the AI images and match the final visuals with what customers will actually receive.
Food images can influence dining decisions. In fact, research from the International Journal of Hospitality Management found that food photos give diners a clearer idea of what to expect than descriptions alone.
For recently added dishes or menu items that need a visual, AI-generated images can be a placeholder while restaurants build their visual library. But it needs careful consideration. A Chicago deli faced backlash for using temporary AI-generated sandwich images that showed misleading food details. Detailed prompts may help produce more accurate images, but restaurants should still make sure the images reflect the actual food they serve.
Start with a photo of the actual dish. AI can improve the lighting, background or setting. Keep the ingredients, portion size, toppings and overall appearance accurate. The idea is similar to traditional food photo editing, where adjustments improve the presentation without changing the actual dish.
As AI images become harder to distinguish, give customers the context of what they’re seeing. Social platforms such as Facebook, LinkedIn and Instagram have done the same. A note such as “AI-generated image for illustrative purposes. Actual dishes may vary.” can provide context. But disclosure doesn’t make an inaccurate image acceptable.
Images should support the menu, not provide facts about the food. Ingredients and allergen information should always come from the restaurant and be checked separately. AI images can change or add visual details, so restaurants must have written information for accurate allergen details.
Menu images can help customers ideate what their food will look like. Whether the image is AI-generated or not, it should look appealing without creating expectations the restaurant can’t meet.
When accuracy matters more than visual polish, AI may be better used behind the scenes. Restaurants can use AI for menu management, analysis, forecasting and other operational tasks where the technology supports the business without shaping customer expectations.
The standard is simple: The food shown on the menu should reflect the food customers expect to receive. AI makes restaurant imagery faster and more affordable, but those savings don’t justify misleading customers.
AI should remove friction for restaurants without removing their control, and appealing doesn’t have to mean fictional.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?