ChatGPT is getting duped by fraudulent storefronts run by scammers — which means you might be getting duped, too.

According to the scam-checking service Ask Silver, the OpenAI chatbot is recommending cloned versions of defunct websites that capitalize on familiar brands to steal your money and bank details, The Guardian reports.

A perfect illustration of the type of scam website that ChatGPT turned up was a knockoff of the British footwear and handbag retailer Russell & Bromley.

“In this instance it looks like scammers are taking advantage of the fact that Russell & Bromley went into administration in January 2026 and was absorbed by Next — so there is no longer an official Russell & Bromley website, but potential customers will likely still be searching for it,” Anna Jones at Ask Silver told The Guardian.

Alarmingly, ChatGPT recommended the fraudulent sites when directly asked for “popular Russell & Bromley purses and bags.” The specific nature of the query might lead to the potential Russell & Bromley customer letting their guard down, expecting the AI to pull from the official website. The fake websites can look convincing and use official-sounding URLs — like “therussellbromleyofficial dot com,” for instance.

Jones told The Guardian that it’s possible that ChatGPT’s underlying LLM has been “poisoned” by malicious content snuck into its training data, which predisposes it to turning up fraudulent pages.

It’s another sign of how even leading AI models can be manipulated to peddle whatever a bad actor wants. Earlier this year, a BBC journalist demonstrated how he’d tricked ChatGPT into telling other users a complete fabrication: that he was a renowned hot dog eating champion, which he accomplished simply by writing a blog post designed to game the chatbot’s algorithms.

With that said, AI chatbots’ proclivity for recommending dodgy websites isn’t new — nor is being used as an avenue for facilitating scams. But it takes another level of urgency as tech companies ramp up their push for letting AI agents shop for you. 

So far, letting an AI pull the trigger on a purchase isn’t nearly as popular as simply using a chatbot for recommendations, but the groundwork for the latter is being laid; ChatGPT lets customers buy everything from within the chatbot interface and has an entire shopping hub, Amazon gives users the option of letting its shopping AI assistant automatically buy products that drop below a certain price, and Google continues to fine-tune a new payments protocol that allows AI agents to make purchases on your behalf. Credit card companies, meanwhile, are working on new technical standards to prevent AI agents from draining your bank account.

If AI agentic shopping hits the mainstream in the future, you can bet that clever scammers will be lying in wait.

“Consumers are increasingly turning to AI tools for advice and recommendations, but criminals are adapting just as quickly,” Louise Baxter, the head of the scams team at the UK’s National Trading Standards, told The Guardian. “The fact that scam websites can appear in AI-generated results is worrying, and is a stark reminder that fraudsters will exploit any new technology that helps them reach potential victims.”

More on AI: Local Brand Realizes Customers Hate Its AI Ads, Switches to Charming Homemade Ones Instead