Scammers are Tricking ChatGPT into Recommending Fake Shopping Websites

Criminals have always been quick to adapt to new technologies. From email and SMS phishing to social media scams and banking fraud, bad actors rarely waste time finding new ways to exploit emerging platforms. The AI era appears to be no different. Recent reports suggest that ChatGPT has, in some cases, directed users to fake shopping websites after scammers successfully manipulated the information available to the chatbot.

Fake stores are showing up in ChatGPT’s shopping results

AI-powered shopping assistants are rapidly becoming a popular way to discover products, compare prices, and find retailers. Instead of opening multiple websites and search engines, consumers can now ask AI chatbots like ChatGPT to recommend a product and provide links to buy it. The promise is simple: faster, easier, and more convenient shopping. But a growing number of reports suggest that these tools may also introduce new risks.

According to a report from The Guardian (via Android Authority), some users have been directed to convincing fake retail websites through AI-generated shopping recommendations. Research shared by scam-detection service Ask Silver reveals that scammers have even created cloned online stores designed to mimic legitimate retailers or established brands, tricking shoppers into handing over money and payment details.

Researchers have linked these scams to a broader tactic known as “AI poisoning.” The concept is relatively simple. Bad actors flood the internet with fake information, cloned websites, fabricated reviews, and misleading content. Over time, some of this material can be indexed, referenced, or surfaced by AI systems attempting to answer user queries. ChatGPT isn’t intentionally promoting scams, but it has been tricked into doing so.

Traditional search engines have spent decades developing systems to identify malicious websites, detect spam, and remove fraudulent results. Even so, scams continue to slip through. AI assistants face an even more complicated task. They don’t simply present a list of websites. They synthesize information and often create a sense of confidence that can make users less likely to question the result.

That’s not necessarily because the AI is more trustworthy than a search engine. It’s because conversational responses can feel more authoritative than a page full of links. As AI shopping features become more elaborate, developers will need stronger mechanisms for verifying retailers, validating sources, and detecting attempts to manipulate recommendation systems. Incidents like these should serve as a wake-up call for the entire industry.