Cloned retail websites are appearing on ChatGPT purchase results, according to scam checking service Ask Silver. The company found that ChatGPT included links to fake versions of footwear retailer Russell & Bromley and home furnishings retailer Dunelm in its responses.
These counterfeit sites accept orders, deliver nothing, and steal payment information. OpenAI has confirmed that fraudulent websites have been removed from its search index and that users can report suspicious sites using the company’s reporting form.
How cloned retail sites are slipping into ChatGPT shopping results
Scam sites imitate the visual design of legitimate retailers and advertise discounts of up to 80%. Their domain names are constructed to look plausible at first glance, using names such as therussellbromleyofficial or russellandbromleylondon.
Ask Silver’s Anna Jones told The Guardian that the AI model may have been “poisoned,” meaning that cloned web pages created specifically to manipulate AI-generated results could have been introduced into the data that ChatGPT extracts from.
The Russell & Bromley case illustrates how the scam exploits real-world events. The retailer went into administration in January 2026 and was absorbed by Next, leaving no official standalone website.
Consumers searching for the brand now have no legitimate independent site to verify, making fraudulent versions easier to present as credible. Jones stated that scammers appear to be directly exploiting this loophole.
Why AI Shopping Results Make This Scam More Effective
AI tools like ChatGPT are designed to reduce friction by summarizing information and presenting direct links as authoritative answers. Users are less likely to examine individual links when the answer is presented as a single, consolidated result rather than a list of search engine entries where the quality of the source is more visible.
Louise Baxter, head of the scam team at National Trading Standards, said: “Consumers are increasingly turning to AI tools for advice and recommendations, but criminals are adapting just as quickly.
“The fact that fraudulent websites can appear in AI-generated results is worrying and a stark reminder that fraudsters will exploit any new technology that helps them reach potential victims.”
Warning signs to look out for and how platforms are responding
Scam sites often share common warning signs despite their seemingly legitimate appearance. Users should be wary of payment requests that request bank transfers instead of standard methods like credit cards or PayPal. Look for domain names that are slightly different from the retailer’s official site.
Deep discounts, often between 50% and 80%, can also be a red flag. Also, check for verifiable return addresses or customer service contacts, which many scam sites lack. Displaying information in an AI-generated format can add a layer of false credibility, making these warning signs easier to miss.
Dunelm has advised customers to only purchase through the company’s official website or authorized app, and confirmed that it takes steps to remove scam sites when it finds them. Russell & Bromley, which was recently acquired, also recognized the situation and has been working to remove the fake sites.
OpenAI confirmed that specific sites flagged by Ask Silver have been removed from ChatGPT’s search index. The company has not provided details on any systemic measures it may be implementing to prevent similar scam sites from appearing in future search results.
What users should do
When using AI tools to research purchasing options, consider embedded links as starting points rather than confirmed destinations. Before purchasing from any retailer mentioned in an AI answer, verify the domain separately with the brand’s official website.
Additionally, check that standard payment options are available and look for clear signs of a returns policy and legitimate contact details.
