We’ve all seen it by now. The distorted omelettes. Melted butter that turns into something else. The unsettlingly literal shrimp toast. Scroll Instagram, TikTok, or Uber Eats and you’ll see a disturbing number of restaurants promoting themselves with AI-generated imagery. Generative AI’s creep into every aspect of life has unsettled me for plenty of reasons. The data centers and the brain rot, for sure, but I have also watched as my wife, an illustrator, loses work as editorial outlets choose AI “art” built off stolen copyrighted imagery, often reasoning human artists are too expensive. Damn those pesky living wages.
Outside my own ethics, AI art at a restaurant just has never made much practical business sense to me. If I’m looking to eat somewhere, the main thing I want to see on that restaurant’s social media (besides the address and hours of operation, for the love of god) is pictures of the actual food they make.
Some restaurateurs argue that using AI-generated imagery saves the money because they can’t afford a marketing team, and platforms like Uber Eats make it easier than ever to implement AI in photo editing. Others say that restaurants have used stock images for ages that don’t actually feature their food, or that food stylists make fast food look too good to be true.
But at the very least, stock images or food styling requires food to exist in the first place. I’m not the only one who finds seeing a crepe that looks like an MC Escher drawing unsettling. Recently, a bagel shop in South Burlington, Vermont faced backlash after posting an ad that used some AI-generated images. Food stylists have accused AI video generators of copying their work. Reddit is full of people trying to figure out if a particular menu image is AI-generated, vowing they won’t eat at that place if it is. “I've found places like this resort to AI images because the actual product made there is not good,” writes Garrett Palm in The Daily Meal. AI slop starts to feel indicative of actual slop.
As more restaurants turn to AI for their advertising, however, they could run up against a new challenge: the legal standard of false advertising. The Federal Trade Commission Act was first passed in 1914 to, among other things, prevent deceptive practices in commerce. It has been used to keep restaurants from claiming they use “local” seafood when they don’t, and against Arby’s for misrepresenting the quality of its roast beef through, yes, food styling. So it certainly seems that posting images of food (or people) that don't exist and claiming it’s what’s on the menu would count as deceptive.
To answer the question, we spoke to Laura Smith, the legal director at Truth in Advertising, a non-profit watchdog founded in 2012 and dedicated to protecting customers from scams and other deceptive practices, whether that’s going after Walmart for deceptive “Made in USA” labels, or filing an FTC complaint against Keurig Dr Pepper over the supposed recyclability of its coffee pods. The Federal Trade Commission is the agency responsible for overseeing any legal violations. Smith anticipates the rise in AI-generated ads will mean a rise in complaints about misleading imagery, and that AI risks pushing the line between embellishment and outright falsehood.
Jaya Saxena: What does Truth in Advertising do, and how do you do it?
Laura Smith: We are a nonprofit, non-partisan consumer advocacy organization and our mission is to eradicate false and deceptive marketing, and to empower consumers to protect themselves against it. We have two major parts to our organization. One is the news reporting side, where we have ad alerts, which educate consumers on a particular ad campaign or a commercial. And then we have the legal side, which I oversee, where we do deeper dives. If there’s a campaign that is especially egregious, or is impacting a particular consumer group that is especially susceptible, we will do a very in-depth investigation. Frequently we present our findings to the company to try to enact change directly. But if we have reason to believe that the company is going to ignore us, or they don't ignore us but they don't make sufficient changes, we then elevate our investigation to what we call a complaint letter that we file with state and or federal regulators. We’ve done that many times, and we've been successful in prompting regulators to turn that complaint into an enforcement action, where they either sue the company in court, or file an administrative action.
When it comes to food and restaurant advertising, what are the issues that most come up?
We get a lot of consumer tips about images that don't look like the food product they received. There are also a number of health and nutrient claims that can accompany food products, and we see if those health claims can be substantiated properly.
But AI can go further because it can depict dishes that don't even exist on the menu, or cuisine styles the restaurant doesn't actually serve, or a level of quality and presentation that has no relationship to the actual product. So now it's taking embellishment and turning it into fabrication.
Are more people bringing up AI imagery in these complaints?
The use of AI and images obviously is an emerging and rapidly developing area, so we anticipate consumer complaints will increase and broaden with that. But the medium doesn't really change the standard. Whether the imagery is a photograph, an illustration, or AI-generated art, the question is the same. And oftentimes the consumers don't know whether it's AI generated or not, if it's really well done. Whether it's deceptive or not, that's the underlying question.
What we’re seeing often is restaurants using AI generated images of food that look nothing like the food they serve. Personally, a local deli is covered in AI art, and the images of the wraps look nothing like what you’d think based on the menu description. It’ll look like a sandwich has tomatoes in it when it actually doesn’t. What is the legal standard here?
The core standard legally is deception under the FTC Act and parallel state consumer protection laws. An ad is unlawful if it's likely to mislead a reasonable consumer in a way that is material to their purchasing decision. This applies to any advertising medium, whether it's TV, print, social media, exterior signage, or promotional apps. So then the question is what do “reasonable consumers” expect with respect to food marketing?
Generally, courts have been skeptical of food imagery false advertising claims, and have dismissed many of them frequently on two grounds. The first is “puffery,” which is a legal term for promotional statements that are so obviously exaggerated and subjective or hyperbolic that no reasonable consumer would take them as literal [Editor’s note: Think of your local diner claiming “World’s Best Coffee!”]. The second ground is the reasonable consumer standard. Courts have come out saying, well, no reasonable consumer would expect that the food item would look exactly like the image. The Supreme Court hasn't weighed in on these food imagery issues, so they've really been deciding on a court by court basis.
With AI-generated art, the question is the same: does it create a false or deceptive impression in the mind of a reasonable consumer that's material to their decision? Traditional food advertising deception involves embellishment, making a real dish look better than it is. But AI can go further because it can depict dishes that don't even exist on the menu, or cuisine styles the restaurant doesn't actually serve, or a level of quality and presentation that has no relationship to the actual product. So now it's taking embellishment and turning it into fabrication. That's where we might see, in the near future, courts coming out a little differently, if the image is too far stretched from the actual product being sold.
It’s definitely become an issue with independent restaurants, who often say they use this imagery to be competitive with bigger chains. Is there a risk that any complaints or legislation is going to affect independent businesses more?
The same law applies whether you're a mom-and-pop restaurant or McDonald's. But for many deceptive marketers, the risk of being caught and facing enforcement action often seems to be just the cost of doing business, and sometimes it’s worth it for them. That might be different for a smaller operation, whether it's worth the gamble of facing the threat of enforcement action, because presumably that kind of enforcement action could have much larger ramifications for a small mom-and-pop restaurant than it would for someone like Arby’s. As to whether they would actually face enforcement action, it really just depends on what the regulators’ resources are like. Sometimes they want to send a message and go after a big name. Sometimes they'll go after the low-hanging fruit.
Given AI imagery is becoming more common, what should consumers be doing to look out for deceptive marketing?
We have a number of ad alerts on food marketing, in particular on the images that are used to advertise food products that don't necessarily match up with the products. We always tell consumers to do their homework, do their research, ask questions. We always say read the fine print. When AI-generated images are used, sometimes there will be a disclaimer somewhere in the ad that the images are AI generated. That doesn't necessarily fix the problem, because the disclaimer has to be clear and understandable to the consumer, but it also has to be conspicuous. And a disclaimer can't cure false ads. But if there's something written somewhere on the ad you should read it, and we also often tell consumers if something looks or sounds too good to be true, it often is.
Consumers are also known to underreport when they feel they've been deceived. Sometimes they think it's not worth the trouble, or they feel guilty or ashamed. Just the fact that there haven't been a ton of complaints about this very specific topic doesn't mean that there's not widespread deception going on.
To report an ad and to learn more about Truth In Advertising, visit the agency’s website.
