AI chatbots are recommending fake retail websites, study finds
The Verge
A new study reveals that popular AI chatbots frequently direct holiday shoppers to fake, risky, or inactive retail websites.
A study conducted by IP protection firm Podqi tested ChatGPT, Gemini, and Perplexity using shopping-related prompts for 151 brands. The research found that all three AI engines direct users to risky websites, which include imposter stores, knockoff sellers, reregistered old domains, and inactive sites.
Perplexity performed the worst in the study, displaying risky sites in 15 percent of its answers. ChatGPT followed in the middle at 8 percent, while Google's Gemini performed the best by showing risky sites in only 4 percent of answers. Additionally, smaller brands were found to be at a higher risk of being impersonated compared to larger companies.
Representatives from Google, OpenAI, and Perplexity responded to the findings, emphasizing their ongoing efforts to improve model accuracy and safety safeguards. Podqi CEO Trevor West advised consumers to verify sources before sharing personal or credit card information.
Why it matters: As more consumers turn to AI chatbots for holiday shopping and product research, directing users to fake websites poses a significant security risk regarding personal and credit card information.





