Google AI Overviews are inventing fake store closures

A Google AI Overview told searchers that Anna Mae’s Bakery and Restaurant in Millbank, Ontario would close for good on July 31, 2026. The business is open and has no plans to close. Google had confused it with an Illinois bakery of the same name that is in fact shutting down, and the false summary sat at the top of search results while customers acted on it. For a single restaurant the damage was a frantic weekend of corrections. For a brand with hundreds or thousands of locations, the same failure mode is a governance problem that scales with the estate.

What happened

According to CBC News, which reported the case on July 23, 2026, the false closure notice surfaced over the weekend of July 18 and 19. Owner Amanda Herrfort found out when a customer sent her a screenshot while she was on her way to a wedding, and she spent the following days correcting the record one phone call and one walk-in at a time. She eventually typed a correction directly into Google’s AI search bar and received an apology and a promise that the error would be fixed. That conversational fix came with no ticket, case number, or service-level commitment she could track.

Google, in a statement to CBC, said the vast majority of AI Overviews meet its bar for helpfulness and accuracy, and attributed the mistake to a gap in reliable information about the topic on the web.

Why it matters

AI Overviews now sit at the top of the results page for a large share of searches, including the “is this place open” and “are they closing” queries that decide whether someone drives to a location today, and Google’s separate AI Mode answers the same questions in its own conversational experience. When the summary is wrong, it is wrong in the most visible position on the page, and it carries the authority of a direct answer that many readers act on without clicking through to the supporting links beneath it. Even on Google’s own account that most Overviews clear its accuracy bar, the errors that slip through cluster exactly where local brands are exposed: facts about named businesses, hours, and open or closed status. Across billions of daily searches, a small error rate is still a large absolute number of wrong answers about real places.

What this means for multi-location brands

The revealing part of Google’s response is that it blamed an absence of high quality information on the web. That is the lever a central team can actually pull. When your open hours, address, permanent-closure status, and business details are authoritative, consistent, and current across Google Business Profile, Apple, and the wider ecosystem, you give the AI a clean source to summarize and less room to fill a void with a guess about a same-named business two thousand kilometres away.

At estate scale this is a monitoring and data-accuracy discipline, not a per-location firefight. Keep local business listings synchronized from one source of truth so every location publishes the same facts everywhere, treat structured, machine-readable location data as the input that feeds AI answers rather than an afterthought, and watch for AI-generated claims about your locations the way you already watch reviews. Positioning your own data as the agentic layer that assistants read from, which is what Places AI is built for, is how a brand stops being at the mercy of what a model infers. The same authoritative-data work sits at the centre of generative engine optimization: the cleaner and more consistent your source data, the more often AI systems repeat it correctly.

“Just like other features in search, issues can arise when there is an absence of high quality information on the web on a particular topic and we use these examples to improve AI overviews broadly.”

Google spokesperson, CBC News

The bottom line

A neighbourhood bakery could fix a false closure with a few phone calls. A national retailer, dealer group, pharmacy network, or charging operator cannot, because the same error can appear against any of its locations at any time and it will not always be caught by a passing customer. The defence is not a complaint filed with Google after the fact, which comes with no guaranteed fix or timeline, but authoritative location data everywhere the AI reads, which gives these systems stronger signals to work from and lowers the risk that they generate a wrong answer about your brand.