Study: Perplexity, ChatGPT and Gemini sometimes send shoppers to fake retail sites

AI chatbots sometimes send shoppers to fake or risky retail websites, according to a study by IP protection firm Podqi shared with The Verge. Across 6,795 answers about where to buy products from 151 brands, Perplexity AI showed a risky site in 15% of answers, OpenAI's ChatGPT in 8% and Google's Gemini in 4%.

Podqi counted fake sites impersonating retailers, knockoff sellers, old brand domains re-registered by someone else and dead domains that others could buy. Its tests ran in September with prompts it wrote itself, such as asking for a brand's regional websites, which triggered risky results most often. Questions about smaller brands surfaced risky sites about twice as often as those about large ones. ChatGPT warned users about a risky site 24% of the time, the most of the three, and in one case Perplexity named the wrong page as Stüssy's official site.

Google said Gemini shows source links so people can check answers, OpenAI said it is improving safeguards, and Perplexity said it never claims to be fully accurate. Podqi sells brand protection services, and one tested brand, Madhappy, is a Podqi client.