AI chatbots linked to risky retail websites, study finds
A study by intellectual-property protection firm Podqi found that AI chatbots sometimes direct shoppers to unsafe retail websites while they research products or look for places to buy them. The analysis covered 6,795 answers from three AI services.
Podqi classified a site as risky if it impersonated a legitimate retailer, sold counterfeit products, used an old brand domain that had been registered by someone else, or was inactive but available for future reuse. Google Gemini had the lowest rate, mentioning or linking to such a site in 4 percent of answers. ChatGPT did so in 8 percent, while Perplexity had the highest rate at 15 percent.
The figures include recommendations, unlabelled displays and warnings about the sites. ChatGPT issued a warning for 24 percent of the risky sites it surfaced, the strongest result among the three services on that measure. Podqi also reported that answers to queries involving larger companies contained risky sites about half as often as answers concerning smaller brands.
Written by Sakhanda Wire AI from reporting by The Verge. Facts, quotes and full context are in the original article below.