← Back to the guide

Guide

What to Do When AI Says Your Business Is Closed

AI answers sometimes tell buyers a business is permanently closed when it is open. Why the error happens, what it costs, and the steps to get it corrected at the source.

By Heather Laskin · Published July 31, 2026

A customer walks in and says they heard you were closing. Then another one asks the same thing. Nobody on your team put out that message, and nothing on your website says it. Somewhere in an AI answer, your business has been marked closed, and buyers have been reading it for weeks.

When an AI answer says your business is closed, it is repeating a wrong or mismatched source rather than inventing the claim. The fix is to correct the underlying records the AI reads, including your business profile, directory listings, and aggregator data, then verify the answer again after a few weeks. Correcting the answer without correcting the source rarely holds.

Why does AI say a business is closed when it is open?

An AI answer is a summary of what the AI found. When it says you are closed, something it read told it so.

Four causes account for most of these errors.

The third cause is more common than owners expect. In July 2026, Anna Mae's Bakery and Restaurant in Millbank, Ontario found that Google's AI Overview was telling searchers it was closing on July 31. The bakery was not closing. An Illinois bakery with the same name was, and the AI had merged the two. Owner Amanda Herrfort described customers arriving mid-season to ask about it: "We have a lot of people coming in talking to our employees saying: 'It says that you are [closing.]'"

For a seasonal business in peak season, that is the worst possible timing.

What does a false closure actually cost?

The damage is hard to see, which is what makes it dangerous.

A buyer who reads that you are closed does not call to confirm. They do not leave a review complaining. They do not show up in your traffic reports as a lost visit. They quietly pick someone else. Your numbers soften and no single cause presents itself.

The Viridian Bookshop in Willoughby, Ohio ran into this in the same month. Google's AI summary described the shop as permanently closed and placed it at 4041 Erie Street when the shop is open at 4081. Co-owner Chanel Bush noticed sales dropping for weeks before a customer mentioned the closure notice. One of her employees, Jay Jackson, put the stakes plainly: "I was so scared because I was worried that we weren't going to be able to stay open."

A shop that opened in October 2025 does not have years of signals working in its favor. A single wrong answer moves the needle much further for a newer business than for an established one.

What should you do first?

Work in this order. The sequence matters more than the speed.

  1. Capture the evidence. Screenshot the answer with the date visible, and note the exact question that produced it. Answers rewrite themselves, and you will want proof of what was said.
  2. Find the source. Ask the AI where the claim came from, and search your business name alongside words like closed, closing, and permanently closed. You are looking for the record that started it.
  3. Correct your own records first. Your business profile, your website, your hours, your address. Make sure your site states clearly and in plain text that you are open, with a visible date.
  4. Correct the second tier. Directory listings, review platforms, industry association pages, aggregator databases, chamber of commerce entries. Every one of these feeds AI answers.
  5. Use the feedback path where one exists. Some AI answers carry a feedback link. Use it, and keep using it.
  6. Recheck on a schedule. Two weeks, four weeks, six weeks. Record what the answer says each time.

Step three is the one owners skip, and it is the one everything else depends on.

Why does correcting the source matter more than correcting the answer?

Because the answer is downstream of the source.

There is no formal ticketing system for AI answer errors the way there is for a listing edit. No case number, no service level, no appeal. Herrfort's route was to type the correction directly into the AI search box, and she got a reply saying the problem would be fixed. Whether a conversational correction persists is not something anyone outside the engine can verify.

What does hold is changing the underlying record. If the closure claim lives in three directories and an aggregator database, fixing those removes the reason the answer exists at all. Fix the source and the answer follows on the next refresh. Fix only the answer and the source regenerates it.

This is the same principle behind any AI visibility work. AI engines assemble answers about your business from sources you mostly do not own, which is why the off-site record deserves as much attention as your own site.

How long does it take to clear?

Plan in weeks, not days.

AI engines refresh their view of the world on their own schedules, and an answer does not change until the sources behind it have been recrawled and reprocessed. Businesses that correct the underlying records generally see answers shift within two to six weeks. Businesses that correct only the visible answer often see the old claim return.

Keep checking through the whole window. An answer that looks fixed on day ten sometimes reverts on day twenty because one stale source was missed.

Who is responsible when AI gets your business wrong?

For a long time the working assumption was that an AI answer was a summary of other people's speech, and the platform simply passed it along. That assumption is now being tested.

On July 24, 2026, the Delaware Superior Court denied Google's motion to dismiss a defamation suit over statements produced by its AI. The court's opinion in Starbuck v. Google LLC put it this way:

The issues presented in this case reveal a new frontier for defamation law, in which artificial intelligence tools are allegedly employed to effectuate the defamatory ends of their makers.

What made the claim survive was scale. The complaint alleged the false statements reached 2.8 million unique users, which supplied the publication element a defamation claim needs. The case now proceeds to discovery, and nothing about it is decided.

For a business owner, the useful takeaway is not that you should sue. Litigation is slow, expensive, and a poor substitute for a working listing. The takeaway is that courts are starting to treat AI output as the platform's own speech rather than a neutral relay. Errors about your business are being taken more seriously than they were a year ago, and documenting them carefully is worth the few minutes it costs.

How do you catch the next one first?

Every business in this article found out the same way: a customer told them, weeks late.

That is the part worth fixing permanently. Once a month, ask the AI engines your customers use a short set of questions about your business. Are you open. What are your hours. Where are you located. What do you sell. Write down what comes back and compare it to last month.

The check takes a few minutes. The businesses that get hurt by a false closure are not the ones with unusual bad luck. They are the ones who were not looking.

If you would rather not run the check by hand every month, a structured AI visibility audit monitors what AI answers say about your business across a calibrated set of questions and flags factual errors before your customers find them. To see what the AI engines say about you today, get a free AI visibility snapshot.

Request an AI Visibility Snapshot →