A study by IP protection firm Podqi found that three major AI chatbots pointed shoppers toward fake or otherwise risky retail websites, with such links appearing in as many as 15 percent of conversations. The test covered Gemini, Perplexity, and ChatGPT and was shared exclusively with The Verge. The findings carry a US consumer angle as the holiday shopping season approaches, when shoppers hunt for in-stock items and the best prices.
Podqi posed shopping-related questions about 151 brands, including fashion labels Lululemon, Birkenstock, Stüssy, and Madhappy, a Podqi client, as well as homeware brand Le Creuset. Prompts included asking for the top websites for buying a brand, where to find a brand online, and whether a brand ships internationally and what its regional sites are. The firm wrote its own prompts rather than drawing on real user conversations, and ran them through a tool that tracks AI answers across chatbots during two testing periods in September.
Across 6,795 answers, all three engines cited and displayed risky sites, which Podqi defined as fake websites impersonating real retailers or selling knockoffs, old brand domains reregistered by someone else, or dead sites that are inactive but could be purchased and reused. Gemini showed risky sites least often, in 4 percent of answers, a figure that includes recommending a risky site, displaying it without comment, and warning about it. Perplexity performed worst at 15 percent, while ChatGPT landed in the middle at 8 percent. ChatGPT warned about risky sites it surfaced most often, doing so 24 percent of the time, and queries about larger companies produced risky sites about half as often as those about smaller brands.
Certain prompts appeared to trigger risky sites more often than others, particularly requests about regional brand websites. The engines at times listed fake regional sites for brands that have none. In one conversation, Perplexity named the wrong page as Stüssy’s official website and described it as the best starting point for guaranteed authentic items.
Podqi CEO and cofounder Trevor West said part of the effort is making sure consumers know that even when a chatbot feeds them a result, they should verify the source before entering personal information or credit card details. Google spokesperson Jessica Levis Boord said the company takes information quality seriously and was happy Gemini performed best among the products tested, adding that Google continually tests and improves its models and displays links to sources alongside responses to help people verify information.
OpenAI spokesperson Taya Christianson said the goal is for ChatGPT to connect people with reliable places to shop, and that the company is continuing to improve safeguards so potentially deceptive sites are less likely to appear in shopping-related results. Perplexity spokesperson Jesse Dwyer said the company never claims to be fully accurate but is relentlessly focused on accuracy.
More AI news from TechManNews.
