Shoppers who rely on AI chatbots to research products or locate retailers are at risk of being directed to fraudulent or deceptive websites. A new study conducted by the intellectual property protection firm Podqi found that popular AI tools frequently suggest risky links, including sites that impersonate legitimate brands or sell counterfeit goods.
Study Findings on AI Shopping Recommendations
Podqi tested ChatGPT, Gemini, and Perplexity by inputting shopping-related queries for 151 different brands, ranging from major companies like Le Creuset to fashion labels such as Lululemon, Birkenstock, and Stüssy. The researchers used a tracking tool to analyze 6,795 responses generated during two testing periods in September. The queries focused on identifying top websites for specific brands, finding online stores, and determining international shipping availability.
The study defined “risky sites” as fake domains impersonating real retailers, sites selling knockoffs, expired domains that have been reregistered by third parties, or inactive sites vulnerable to being purchased for malicious use. According to the data, all three AI engines directed users to these risky sites to varying degrees:
- Perplexity: Performed the worst, with 15 percent of its answers containing risky sites.
- ChatGPT: Included risky sites in 8 percent of its responses, though it provided warnings about the sites 24 percent of the time.
- Google Gemini: Performed the best among the three, linking to or mentioning a risky site in 4 percent of its answers.
The research also noted that smaller brands are at a higher risk of being associated with fraudulent sites, as queries for larger companies resulted in risky links about half as often as those for smaller ones.
Risks for Consumers and Industry Responses
As AI companies continue to promote their chatbots as tools for product research and discovery, the potential for consumer harm grows, particularly ahead of the holiday shopping season. Experts warn that users may be tempted to trust AI-generated results without verifying the legitimacy of the websites provided.
“Part of this piece, too, is making sure that the consumers know, 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,” Podqi CEO and cofounder Trevor West told The Verge.
Representatives from the AI companies responded to the findings by emphasizing their ongoing efforts to improve accuracy and safety. A Google spokesperson, Jessica Levis Boord, stated that the company takes information quality seriously and noted that Gemini displays links to sources to assist users in verifying information. OpenAI spokesperson Taya Christianson said the company aims to connect users with reliable shopping destinations and is working to implement safeguards against deceptive sites.
Perplexity spokesperson Jesse Dwyer defended the platform’s performance, noting that the company is “relentlessly focused” on accuracy while acknowledging that it does not claim to be perfect. The study highlighted that certain prompts, particularly those asking for regional brand websites, were more likely to trigger the inclusion of fake or incorrect links, with Perplexity in one instance incorrectly identifying a page as the official site for Stüssy.
