When you want to find a new local business, who do you ask? Today, 45% of Americans are turning to an AI assistant to get recommendations for local businesses — up from 6% a year ago, according to BrightLocal's 2026 Local Consumer Review Survey. It's a dramatic change in user behavior and shows that many people are quickly adopting this new habit. However, the data that AI assistants use relies on star ratings, among other sources. And as Nextdoor reports below, researchers and experts say star ratings should not be fully trusted.
Federal Trade Commission (FTC) economist Devesh Raval has studied ratings across major review platforms. According to his 2024 research published with the FTC, most businesses average four stars or better, including those that have real complaints on file and poor ratings. Almost a third of reviews were found to be just a number, with no additional text to bolster the review and ground it in context. Almost no rating falls below two stars. In other words, the difference between a great plumber and one who could inadvertently flood your kitchen could be a tenth of a point that doesn't measure much at all.
Regulators have noticed how easily reviews can be faked. In 2024, the FTC finalized a new rule banning fake and bought reviews, as well as reviews written by people who have a material connection to a company but don't disclose that connection. Then, in December 2025, the agency started sending warning letters to companies over fake and pay-for-play reviews. Each violation carries a penalty of $53,000 or more. It's too early to tell whether these attempts are working.
Last December's letters were the first act of enforcement from the FTC. That leaves a gap at an awkward moment. More people are asking AI assistants for help finding a pediatrician or a dog walker, and those assistants can be steered. In May, a BBC reporter published a made-up claim about himself on his personal website, and within a day, several major AI tools, including ChatGPT and Google's AI Overviews, were repeating it as fact. Google has since updated its spam policies to address the problem. An assistant that leans on the same lopsided ratings the FTC economist found will hand back an unreliable answer.
Shoppers don’t trust AI summaries by themselves
People may be discovering businesses with AI assistants, but a single AI summary still gives them pause.
Among people who are using AI-generated review summaries to make purchase decisions, 82% go beyond ChatGPT or Claude, according to the BrightLocal survey. Most are visiting websites, social media feeds and other review platforms before they book a service. In fact, just 23% of consumers make decisions solely from reading an AI summary, suggesting that AI is just the first step in the shopping process, not a decision-maker.
Alternatively, trust in reviews remains high. The BrightLocal survey found that 49% of people say they place as much trust in an online review as they do a recommendation they’d get from a friend. However, word of mouth continues to outweigh reviews from an unverifiable source. If someone you know has had a bad experience with a business, say an auto mechanic that messed up and kept them from picking up their kid from soccer practice, that anecdote is very likely to shift your opinion of said business. The FTC's research suggests that the numbers making up many online reviews haven’t earned that trust.
Local information is still relevant, research says
Getting a recommendation from someone who can be identified and lives near you can carry a cost for that person if it turns out to be wrong. An anonymous five-star rating costs its author nothing, whether the writer is a genuine customer or a paid employee of the business. That difference can push people toward their local circles. An AI summary flattens it. It reports an average and leaves out who wrote each review and whether that person lives down the street or across the country.
Neighborhood-level sources matter to the people in those communities. In 2024, the Pew Research Center's research found that 52% of U.S. adults at least sometimes get information about the places they live from community-centric online platforms like Facebook groups.
Maintain skepticism for reviews fetched by AI
The written context of a review says more about it than the number of stars associated with it. Specific information, like the cost of a particular service or details about the problem an electrician solved, can give consumers more to go on. What a star leaves out, neighbors, friends, and coworkers can fill in.
Today, regulators are treating fake reviews as a legitimate legal issue. As AI assistants become a first stop for local recommendations, the ratings they draw on deserve the same scrutiny. Research shows most people are still thinking twice about what an algorithm shows them about a business. For now, the recommendations that are still worth the most come from someone who’s willing to put their name on them.
This story was produced by Nextdoor and reviewed and distributed by Stacker.
