When AI sends your customers to a fake store: what a brand can do

Yes, an AI assistant can send your customers to a fake copy of your store. In June 2026 the UK scam-checking service Ask Silver asked ChatGPT about popular Russell & Bromley bags, and some of the source links in the answer led to fake copies of the retailer's site, advertising discounts of up to 80%. It found a cloned Dunelm site in ChatGPT's answers too.

OpenAI, Next and Dunelm each responded to what Ask Silver found, and each response dealt with fakes that were already live. That is worth thinking about if you sell online, because how soon you know a fake copy of your store exists, and how long it survives, is the one part of this you control.

If a copy of your store is already live, start with what to do when someone has cloned your website, or report it through our urgent help page.

What happened in June, and why it matters to your store

OpenAI said it had removed the sites Ask Silver found from its search index. Next, which bought the Russell & Bromley brand in January, said it was working to have them taken down. Dunelm told customers to use only its official website or app, and said it works to have fake sites removed as quickly as possible once it knows about them (Retail Gazette, 8 June 2026).

The sites Ask Silver found were built to look like the official retailer's, on official-looking web addresses, with prices low enough to make a shopper hurry. A fake store exists to take payment details. The coverage we have read describes no shopper who lost money to these particular sites. They came to light because a scam checker went looking.

The Russell & Bromley case had a detail that matters to any brand. The original business stopped trading after administrators were appointed on 21 January 2026, and Next bought the brand. Ask Silver's Anna Jones said, as Digital Trends reported it, that with no official site remaining after Next absorbed the brand, fraudsters built a convincing clone, and that shoppers were still likely to search for Russell & Bromley directly. A well-known name with nowhere official to land is a gap, and a fake store can fill it.

Why the links appeared in ChatGPT's answers has not been established, and knowing wouldn't change what a brand can do about it. In a report published in May 2026, F-Secure researcher Dr Megan Squire argued more generally that “weak merchant verification and limited safeguards may allow fraudulent stores to gain visibility”.

National Trading Standards warned shoppers not to assume a site is genuine just because it appears in an AI-generated answer. “Consumers are increasingly turning to AI tools for advice and recommendations,” said its Louise Baxter, “but criminals are adapting just as quickly.” A shopper who asked a question and got a confident answer has fewer reasons to look twice.

What a fake store costs the brand whose name is on it

When a shopper does pay a fake store, they lose the money first. The rest lands on the real brand, and none of it arrives as a line in your accounts.

• The sale. Shoppers search for brands they know by name, as Ask Silver noted of Russell & Bromley, so some of the people who find a fake were looking for you. Each one who orders from the copy is an order you didn't take, and a shopper who thinks they have already bought has stopped shopping.
• The complaint. If the order goes wrong, the customer goes looking for the brand whose name was on the site. Their bank's dispute goes to the fake. Their complaint comes to you.
• Your team's time. Your customer service staff answer the “where is my order?” emails for orders you never received, and have to tell a customer, kindly, that they were scammed in your name. Someone else on the team is meanwhile trying to get the fake taken down.
• Trust. A customer taken in once is warier next time, including when your genuine sale email arrives. If you sell through resellers, a fake at 80% off undercuts them too.

How to stop AI assistants sending your customers to fake copies of your store

You can't edit an assistant's answers. You can make your real shop easy to recognise, find the copies early, and shorten the life of each one. Once a fake store is offline, a link to it in an AI answer leads nowhere.

1. Ask the assistants where to buy your products

Do what Ask Silver did. Ask the AI assistants your customers use where to buy your best-known products, and whether any are on sale. Open every source link and read the address. Then search for your brand name alongside words a fake store might use, such as “sale”, “outlet” or “official”. Repeat the check after launches, during sales and whenever your name is in the news.

2. Make your official shop unmistakable

Say on your own site which web addresses you sell from, and list your authorised stockists if you have them, so anyone checking has an official answer to compare against. If the brand changes hands, relaunches or moves to a new domain, keep the old address working and pointing to the new shop. Don't leave the kind of gap Ask Silver described.

3. Tell customers how to check

Give customers one page with your official shop addresses, what a genuine offer from you looks like, and where to send anything that looks like a copy.

4. Watch every day for new copies

One check finds today's fakes. New ones keep appearing: on addresses that put an extra word next to your brand name, on domain endings you never registered, and sometimes on addresses that have nothing to do with your name at all. Finding them while they are new means checking new domain registrations and searching the web for copies of your pages every day. By hand, that is a daily task, and the kind that gets pushed to tomorrow when the shop is busy.

5. Get each fake taken down, and keep checking it has gone

This is the hard part. A fake store relies on several companies: the one its domain was registered through, the one hosting it, and often the store platform it was built on. Each can take it offline once it is impersonating you, and each has its own way to report abuse and wants the evidence in its own form (see which lookalike domains can be taken down). Some reports come back pointing you to someone else. Others get no reply. Every one needs following up, and somebody has to keep going back to check the shop has really gone. When the same store reappears on a new address, you start again.

Report it to the company behind the assistant as well. OpenAI said it had removed the Russell & Bromley fakes from its search index. That takes the fake out of one assistant's answers, and the site stays live for everyone who reaches it another way.

In many e-commerce teams all of this lands on whoever found the fake, usually someone whose real job is running the shop. Some brands hand it to their lawyers instead.

Or hand steps 4 and 5 to us. DefendDomain keeps watch, day and night, for lookalikes of your shop's domain and for copies of your store, even on web addresses with no trace of your name. When a fake store goes live, one click starts its takedown. The reports, every follow-up and every check then run for you, with updates on where it stands, and on our fully managed plan we make the click too. Our requests also go to search engines and browser warning services, so a copy can be delisted and flagged while it is being taken offline. Takedowns are unlimited on every plan. More on how we find and remove fake stores that use your brand.

For an upmarket US e-commerce brand that sells through its own store and a network of partners and resellers, we found seven fake stores in its first three days with us, the first within nine hours. Five domains were taken down in its first week. Its takedowns used to go through its lawyers. DefendDomain runs them now.

Check your domain for free, or book a demo to see what we find for your store.

For the work to do before any fake appears, such as registering the obvious variations of your name, see our guide to preventing brand impersonation.

Frequently asked questions