Perplexity, ChatGPT, Gemini pitch shoppers risky retail sites
TL;DR
- Podqi found Perplexity surfaced a risky shopping site in 15% of answers, ChatGPT in 8%, and Gemini in 4%, across 6,795 answers.
- In 57% of answers that included a risky site, the chatbot actively recommended it; risky sites appeared 6.3 times more than warnings did.
- Smaller brands took roughly double the risky-link exposure of large ones; ChatGPT flagged a risky site as unreliable in only 24% of cases.
Perplexity recommended a risky shopping website in 15 percent of its answers, ChatGPT in 8 percent, and Gemini in 4 percent, according to a study by IP protection firm Podqi shared with The Verge. Across 6,795 answers about 151 brands tested in September, Podqi flagged 718 risky domains, 351 of them fake and 367 dead.
"Every chatbot surfaced risky destinations. The rates differed considerably," the report states. In 57 percent of answers that included a risky site, the chatbot actively suggested it as a place to shop. Risky sites were recommended 6.3 times as often as they were warned against.
Smaller brands took roughly double the exposure of large ones. Queries about regional storefronts fared worst, in part because some of those regional sites simply do not exist. In one case, Perplexity named an incorrect page as Stüssy's official site and described it as a source for authentic products.
When ChatGPT did surface a risky link, it flagged the site as potentially unreliable in only 24 percent of those cases.
Podqi CEO Trevor West advised shoppers to verify any source before entering personal or payment information, to go straight to a brand's own site, and to watch for misspelled domains and odd extensions. The study warns that "a shopper who follows an AI-generated link to an impersonator site risks handing over payment details, login credentials, or personal information."
All three companies responded to Podqi's findings. None disputed the core finding.
Shared on Bluesky by 2 AI experts
-
A study testing Gemini, Perplexity, and ChatGPT found that in as many as 15 percent of conversations, chatbots displayed ‘risky’ URLs. www.theverge.com/ai-artificia...
View on Bluesky → -
Exclusive: Ahead of the holiday shopping season, a series of tests found that Gemini, ChatGPT, and Perplexity are recommending fake and other risky retail websites when asked where to buy things online. www.theverge.com…
View on Bluesky →
Originally reported by theverge.com
Read the original article →Original headline: AI chatbots are recommending fake retail websites, study finds