AI
Study reveals AI chatbots frequently point shoppers to counterfeit retail sites
A new analysis found that Gemini, ChatGPT, and Perplexity regularly recommend risky websites when users ask for shopping advice, creating potential security risks as the holiday season approaches.
1/2
The Verge
Consumers relying on AI chatbots to locate retailers and compare products face a significant hazard: the systems frequently direct them toward counterfeit storefronts and other deceptive web addresses. A fresh investigation has exposed this vulnerability across three major platforms, raising fresh concerns as holiday shopping season nears.
How the research unfolded#
Intellectual property firm Podqi evaluated ChatGPT, Gemini, and Perplexity by posing straightforward shopping queries about 151 brands across fashion, homeware, and other sectors. The team asked questions such as "What are the top websites for buying [brand]?" and "Where can I find [brand] online?" over two test cycles in September. Across nearly 6,800 chatbot responses, Podqi identified instances where systems recommended what they classified as "risky" URLs—fraudulent impersonations of legitimate retailers, knockoff peddlers, expired domains seized by malicious actors, or defunct sites ripe for reregistration.
Performance gaps between platforms#
The three systems diverged markedly in their accuracy. Google's Gemini demonstrated the strongest safeguards, surfacing problematic sites in just 4 percent of responses. OpenAI's ChatGPT landed in the middle at 8 percent, though it issued cautions about suspect links in roughly a quarter of instances where it did flag them. Perplexity lagged significantly, with concerning websites appearing in 15 percent of all answers. Smaller brands proved particularly vulnerable to misrepresentation, with risky site recommendations appearing roughly twice as often for lesser-known companies compared to household names.
Why this matters now#
As major retailers prepare for increased holiday traffic, the stakes grow higher. More shoppers will likely turn to chatbots for product discovery and purchasing guidance, especially when hunting items from unfamiliar brands or seeking competitive pricing. Someone relying on AI guidance might unwittingly enter payment or personal data into a fraudulent page, potentially exposing themselves to fraud or identity theft.
Podqi's CEO emphasized the importance of user vigilance: "Even though you are being fed a result in a particular chatbot, it is always best to verify the source of that before entering personal information or credit card information."
Company responses#
Google acknowledged the findings in a statement, with spokesperson Jessica Levis Boord noting that Gemini "performed best among the products tested" and pledging continued refinement. OpenAI indicated it is "continuing to improve our safeguards to make potentially deceptive sites less likely to appear in shopping-related results." Perplexity defended its track record on accuracy benchmarks but did not specifically address the study's shopping-related findings.

