Damian Mansell was shocked to discover what Google had to say about his fledgling business. When he searched for reviews of The Plastics Shed, an online supplier of building materials registered in early 2025, the platform’s AI stated that customer feedback was “predominantly negative.” Fortunately for Mansell, these reviews did not pertain to his company. However, resolving the issue proved to be a challenging process.
Complaints included delayed deliveries, dishonest staff, and damaged products. Although the Google AI summary noted some positive reviews, it indicated that poor customer service was a “significant, recurring problem.” It turned out that these reviews were actually about competitors and businesses selling literal plastic sheds. Mansell was left unsure of how to address the situation.
“I understand how this happens, but why is it occurring on Google? I mean, this is top AI technology,” said Mansell, who resides in the UK.
Adding to the dilemma, Mansell was paying Google approximately £700 (around $850) monthly for advertising, while these summaries discouraged potential customers from his business. Following advice from internet forums, he reported the erroneous summary to Google multiple times. Eventually, after several weeks, the situation began to improve.
While he finds AI to be an invaluable tool for business growth, Mansell wishes for greater accountability when things go wrong. “AI provides ordinary people with knowledge, but it can also ruin someone’s life,” he remarked.
Google’s AI reviews swiftly become the first impression for consumers regarding businesses. Instead of browsing through reviews and websites, many users encounter a single summary that claims to aggregate information from across the internet. When such summaries are inaccurate, misleading, or chaotic, business owners assert that they can cause significant reputational harm and financial losses, with limited options for recourse.
Mansell continues to ponder how much revenue this issue has cost him. He attempted to persuade Google to refund part of his advertising spend, but despite his efforts, he has been unsuccessful.
“Once I latch onto something, I don’t let go,” he stated. “But with Google, I’ve encountered a formidable opponent.”
Online searches have transformed significantly in recent years; instead of a list of links, Google frequently provides a single answer that summarizes the entire body of information available online. These reviews appear credible and sound confident but rely on a long list of sources with varying reliability. They can also present outright false information, such as suggesting that people make pizza with glue.
While misinformation online is not new, the concise package delivered by AI is a novel development.
“In traditional search engines, users see multiple results. If one is incorrect, there’s a chance that something else will balance it out,” said Chirag Shah, a professor at the Information School at the University of Washington.
This balance has vanished; with AI summaries, users receive a “response” that may be correct or incorrect.
Google AI Confuses Businesses
Earlier this year, Betty Whitney began noticing that Google reviews were confusing her business, NW Select Property Management in Idaho, with others that had similar names. The top panel appeared to merge her company with one in the area that closed years ago, mixing her reviews with those of property managers from other states.
“If I were a customer looking for a property management company and saw an AI review, I wouldn’t call myself,” Whitney explained.
For months, Whitney has been trying to rectify this situation. Like Mansell, she utilized Google’s review mechanism, leaving negative ratings when they were erroneous, and made updates to her website to provide more accurate data to AI. After describing her issue on a Google help forum, she contacted an external SEO expert, who was able to assist her—at least partially. The review improved but still occasionally reverted to the erroneous version.
Real Estate Agent Struggles with Google
Three years after moving, Philippa Main, a real estate agent in Northern Virginia, still can’t convince Google that she is no longer in Florida. When she searches her name, most of the information displayed is correct, but then she sees a statement confidently declaring she has “served the Tampa Bay area since 2014,” despite being located just above her Virginia address.
Main has tried everything she could think of to correct this: combing through the internet to find out where AI sourced this information, writing to Google, and asking friends to report the problem.
“In my industry, competition is so fierce that if anything seems off, someone will just call the next person on the list,” she noted.
She added that this is particularly frustrating for small businesses, as “we’re just trying to do everything we can to compete with these huge firms that actually have direct contact with Google or its representatives.”
“Google just seems indifferent,” she said.
Google Responds
In a statement, a Google spokesperson said the experiences related to AI in search are “rooted in quality ranking systems and aim to present diverse perspectives” from across the web.
It’s no secret that AI is not always a bastion of truth—almost anyone who has used this technology has, at some point, received a response that diverges entirely from reality. As reported in April by the New York Times, Google processes over five trillion searches annually, and even if its reviews are correct nine times out of ten, that still amounts to half a billion errors.
Google acknowledges that while the vast majority of its reviews are accurate, there are instances where context is lacking or content is misinterpreted. The spokesperson also stated that the study referenced by the Times has “serious gaps.”
Google’s summaries synthesize information from multiple sources—its website itself is a major source, but the platform also absorbs posts from Reddit, decade-old blogs, and reviews from Yelp. While the platform provides links that are supposed to verify its claims, these links do not always support the generated response. Furthermore, AI is known to hallucinate, fabricating seemingly credible information out of thin air.
AI Review System Leaves Room for Error
“There is significant room for error here,” said Lily Ray, an SEO and AI search consultant and founder of the consulting firm Algorythmic.
These reviews are presented with such confidence that people are unaware they are looking at a distilled or generated answer that may be incomplete or incorrect. Instead of clicking five links to compare information or spending a few minutes confirming it, users quickly scan the automated summary and consider the matter resolved. If AI indicates that a roofing company has terrible reviews, they move on to the next one. Few take the time to check whether complaints are about a different business entirely.
Search industry specialists state that this marks a new era for businesses. They no longer need to focus primarily on search result rankings, but rather manage how AI interprets their reputation. It’s no longer about chasing clicks—it’s about ensuring that AI knows you exist and speaks of you positively and accurately.
Ray describes this as “the biggest change in search” she has seen in her 16-year career.
This brave new world brings a host of nuances and complications. Google is quick to cite Reddit, which, as Ray points out, “can go wrong very quickly.” While there is much valuable information on Reddit, it can also be quite wild. The same applies to comments on YouTube, which AI seemingly favors as well.
Some brands suffer from an information void: little is written about them, so AI tries to fill the gaps or invents incorrect answers. Alternatively, they may have issues with their name being too similar to another entity, causing the model to struggle to distinguish who is who. There can also be individuals acting in bad faith who intentionally leave false or negative information about businesses online, which AI summaries then pick up. Even basic facts like store hours or phone numbers require coordinated effort across the internet to maintain accuracy.
How to Prevent AI from Confusing Businesses
“Maintaining content and brand information accurately and up to date requires tremendous effort,” Ray stated.
Michael King, founder and CEO of the digital marketing agency iPullRank, emphasizes accuracy in citations and gaining some influence over AI responses.
“These systems work by effectively conducting numerous searches in the background and then feeding content to a large language model,” he explained.
Businesses need to create a larger “surface area” by publishing more content and targeting more keywords to help AI find the right answer. He encourages clients to position themselves as experts in their own brands.
“It’s more about thinking of it as a reputation management campaign than a traditional SEO campaign,” King remarked. “It’s significantly more multidimensional.”
Ben Fisher, who works in this field, still encounters issues. He noticed that Google’s AI summary warned that his company—Steady Demand, a social media and SEO consultant for small businesses—is a scam. Upon further investigation, he discovered it was referencing an old Reddit thread regarding an app with a similar name, and he had to spend time “teaching” Google how to differentiate between the entities.
“The biggest problem is that there’s no one to contact. The second major issue is that you search once and that’s it,” Fisher said.
Large language models do not always produce the same results, even for the same queries, so users may not realize that one result can differ from what they’ll see a month, week, or minute later.
“It remains a situation where you need to monitor everything regularly,” he added. “Otherwise, you simply won’t know why no one is calling.”
This is a complex issue from technological, entrepreneurial, and regulatory perspectives. LLMs are improving, but they will never be perfect. Business owners can do their best to manage how they are represented in search results, but they also have 9,000 other things to do. Reasonable people—and various countries and legal systems—may disagree on how much responsibility Google should bear when a bot makes a mistake.
Legal Action Against Google?
In Canada, a musician filed a lawsuit against Google for $1.5 million, claiming that an AI summary falsely identified him as a sexual offender. A court in Germany recently issued a preliminary ruling stating that Google is liable for false statements included in its AI reviews.
The Google spokesperson noted that the case in Germany pertains to “specific and narrow errors,” not the way reviews present content, and the company disagrees with the ruling and plans to appeal.
Shah stated that in the United States, consumer protection in such cases is “very limited.”
“I don’t think lawmakers fully understand this technology and its implications enough to do anything,” he said.
Meanwhile, business owners are left to grit their teeth and hope that the mysterious technology behind these AI summaries will treat them kindly. This is the case for Mansell, who proudly states that Google’s review of The Plastics Shed is now “fantastic,” as are many of its genuine reviews.
“What worries me is what can be said about you without any chance for recourse,” he reflected.
Despite the frustration over the summary and failed attempts to obtain a refund, Mansell continues to pay for Google ads—otherwise, people do not visit his website.
“I’ve just given up,” he said. “It was a completely pointless endeavor.”
