Is AI sending shoppers to a fake copy of your store?
November 14, 2026 · Clever Bean
DRAFTED OVERNIGHT. APPROVED AND PUBLISHED BY DARIA BYKOVA. November 14, 2026
Being named by an AI chatbot proves nothing until the link beside your name is checked, because the address is where a shopper's money goes.
Ask each assistant a fixed set of questions about your store, then check every answer for three things: the web address it gives, whether it calls you open, and whether prices, hours and contact details match your site. Save the answers with dates and repeat the set on a schedule so you can see change.
What did the 13 November 2026 report find about chatbots and fake shops?
The news reached this page on 13 November 2026, so read this section as dated. Digital Today reported that the intellectual property protection firm Podqi had reviewed ChatGPT, Gemini and Perplexity. Risky shopping sites showed up in about 9 percent of 6,795 responses. The risky sites included fake shops impersonating brands, counterfeit sellers, former brand domains re-registered by third parties and websites that had stopped operating. In total the investigation named 718 risky domains: 351 fake and 367 inactive.
The report adds that Google pointed out Gemini recorded the lowest exposure of the three.
Our opinion, in plain sentences. The percentages will change as the assistants change, so do not build a plan around 9. The two types of risk are the useful part: a copy that pretends to be you, and a dead address that once was you. Both can happen to a store that did nothing wrong, for example after a domain change or a lapsed registration. The test took place in September, it used prompts the researchers wrote, and it measured what appeared in answers, not what shoppers did next. Treat it as a reason to look at your own answers, because your own answers are the only ones that matter to your sales.
Risky domains found in the Podqi review of chatbot shopping answers
Source: AI chatbot shopping recommendations include risky sites, hundreds identified (Digital Today), 2026-11
The review named 718 risky domains across ChatGPT, Gemini and Perplexity answers: 351 fake sites and 367 inactive ones. Dead addresses were as common as fakes, so a check for a lookalike is only half the job. Check for a closed or abandoned version of your store too.
How do I find out what ChatGPT, Perplexity or Gemini say about my store?
Ask them. Open a fresh session in each assistant, ideally logged out or in a private window, so earlier chats do not shape the answer. Type your questions as a shopper would, copy the full answer and every link into a document, and add the date and the assistant's name.
Nobody can read what other people type into ChatGPT, so you cannot see the real queries about your store. You can only sample answers. Answers also differ from run to run, so ask each question two or three times before you call something a pattern.
What search queries should I run in AI chatbots to test my store?
Use questions at four levels: who you are, where to buy, whether you are open and what things cost. Replace the brackets with your own details.
Test prompts to run in each assistant
0 of 8 done
Details
Checks the description. Look for wrong product ranges, wrong country or a different business with a similar name.
Details
Checks the address. Compare the domain letter by letter with yours, including the ending (.com, .shop, a country code).
Details
Checks whether the assistant lists you, resellers or a copy. Open every link.
Details
Checks status. Any answer that says closed, moved or rebranded needs a source.
Details
Checks hours and contact details against your site and your business listing.
Details
Checks prices. Note whether the price is current, old or from a reseller.
Details
Regional questions are where invented sites are most likely to appear, so test every market you do not serve with a local site.
Details
Checks what trust signals the assistant repeats and which pages it cites.
Run eight questions in each assistant: description, official website, where to buy, open or closed, hours and phone, price, regional site and legitimacy. Open every link and compare the domain with yours. Record answers with dates.
How can I tell if a chatbot is linking shoppers to a fake or lookalike store?
Start with the address. Copy each link from the answer and compare it with your domain. Lookalikes tend to add a word (store, outlet, sale), swap an ending or change one letter. A chatbot often shows a brand name next to a link, so a correct name beside a wrong address is the signature of this problem.
Then open the page in a private window and look at it as a shopper. Do the product photos come from your site? Are prices far below yours? Does checkout ask for payment details by bank transfer or on an unfamiliar processor? Does the contact page show an address and name you do not recognise? Any yes is enough to take screenshots and start the report steps below.
Also search your store name in Google and look at the first two pages for domains you do not own.
How do I check whether an AI assistant shows my store as closed or out of business?
Ask the status question directly and also ask it indirectly, for example "where can I buy [product] from [store name]". The first reveals a statement and the second reveals what the assistant leaves out.
If an answer says you are closed, find the page it cites. The usual causes are an old directory listing, an article about a past closure or a former domain that no longer loads. A domain you let lapse can also be re-registered by someone else, which is one of the cases the report above named. Type your old domains into a browser. If one now shows something you did not publish, treat it as a lookalike.
How exposed is your store to wrong AI answers?
Question 1 of 4
Have you ever searched for your store in ChatGPT, Perplexity and Gemini and opened every link?
The quiz scores four things: whether you have looked at your answers, whether an old domain or location change could mislead an assistant, whether your details match across listings, and whether you sell in markets without a local site. Higher scores mean more gaps. The verdicts end in one step.
What should I do if a chatbot gives wrong prices, hours or contact details?
Assistants build answers from pages they can read, so the fix usually sits on the source page, not inside the chatbot. Work in this order:
- Find the page the answer cites or paraphrases. Perplexity and Gemini show source links; for ChatGPT, ask "what source did you use?".
- If it is your page, correct it and make the fact easy to read: price in plain text on the product page, hours and phone on a contact page.
- If it is a directory, marketplace or business listing, correct it there. Give every place the same name, address, phone and hours.
- If it is a third-party article, ask the publisher to update it.
- Use the thumbs-down or feedback control in the assistant to flag the answer, and note the date.
- Ask the same question again in a week and log the result.
Corrections take time to show. Nobody can promise a date when an assistant will repeat the new fact.
How do I report a fake store or wrong listing to an AI platform?
There are two targets, and the second matters more. The first is the assistant: use its feedback control on the answer and attach the link and a short description. Button names change, so look in the assistant's help pages for the current route.
The second is the fake site itself. Take screenshots with the date, note the domain and look up who hosts it and who registered it. Send a report to the hosting provider and the registrar, using the abuse contact that each lists. If the copy uses your photos or name, say so in the report and include your original pages. If it takes payments, tell the payment provider shown at checkout. If your brand name is a registered trademark, the trademark route with the registrar can be faster, and a lawyer can advise on that for your country.
Keep every report and reply in one folder. A pattern of repeats helps when a case stalls.
How often should I re-check AI chatbot answers about my store?
There is no universal schedule, because stores differ. Choose a rhythm from your own risk. A store that changed domain, moved or runs seasonal peaks needs closer checks than one with a stable site and a single listing. A sensible starting point is a full run of your prompts monthly and a quick run of the address and status questions before any busy season or after a site change.
Use the model below to see how much reading your own schedule creates.
How long does your re-check routine take? (a model with your inputs)
- Answers to read each month
- 24
- Hours per month
- 1.2
This is a time model, not a measured benchmark. Minutes per answer depend on how many links you open.
With the default of 8 questions, 3 assistants, one round a month and 3 minutes per answer, you read 24 answers and spend about 1.2 hours a month. Double the rounds and the time doubles. Change the sliders to match your own routine.
Which tools can monitor brand mentions and AI answers automatically?
Three kinds of tool cover this, and you can combine them.
- Free alerts and listings. Search alerts for your brand name and a monthly look at your business listing catch new pages that mention you.
- Domain watching. Many registrars and security services can alert you when a domain similar to yours is registered. Ask yours what it offers.
- AI answer tracking. Some paid services run a set of prompts on a schedule and log what each assistant says. Compare what they cover (which assistants, which countries, how often) before paying, and check that you can see the full text of each answer, not only a score.
A spreadsheet with the date, assistant, question, answer, link and verdict column does the same job for one store and costs nothing but time. Automate it once the manual log shows you what you are looking for.
Can I see what shoppers actually ask ChatGPT about my store?
No. Nobody outside the assistant's provider can read live queries. You can sample answers with test questions and watch your own site analytics for visits from assistants.
Why does the same question give different answers?
Assistants generate each answer fresh and may use different sources each time. Ask each question two or three times and look for what repeats.
Is a wrong answer always the chatbot's fault?
Often the source is an old listing, an outdated article or a conflict between your own pages. Fix the source first, then flag the answer.
What if the fake store uses my product photos?
Save dated screenshots of the copy and your originals, then report to the host and registrar and to any payment provider at its checkout.
Do I need to check every assistant?
Test the ones your buyers use. Three to five is enough to see a pattern, and the answers differ between them.
- AI chatbot shopping recommendations include risky sites, hundreds identified (Digital Today) 2026-11
Chatbots will keep describing your store, and some of those descriptions will be wrong. A fixed question set, a dated log and a rule for who fixes what turn that into a routine job of under an hour a month. If you want a first read before building your own log, the free check shows how your business appears in Google and in five AI assistants, with three things to fix first: https://cleverbean.ai/#check