Artificial Intelligence

Fake Shopping Sites in AI Answers: 9% of Chatbot Replies Risky

A Podqi study says AI chatbots sent shoppers to fake shopping sites in about 9% of answers. The numbers, the limits and how brands can protect themselves.

AI chatbots recommending fake shopping sites study
In this article
  1. What the Podqi study tested
  2. How each AI chatbot performed
  3. What Google, OpenAI and Perplexity said
  4. Limits of the fake shopping sites data
  5. What brands should do now
  6. Why smaller brands should act first

AI chatbots are sending some shoppers to fake shopping sites, and smaller brands are hit hardest. That is the finding of a study by Podqi, an IP protection company, which tested ChatGPT, Google Gemini and Perplexity with shopping questions about 151 brands. The Verge first reported the results on November 12, 2026.

Across 6,795 answers, about 9% exposed the shopper to a risky website. That includes sites impersonating a retailer, sites selling counterfeits, abandoned domains taken over by someone else, and dead domains that a fraudster could register.

For marketers, this is a brand problem as much as a security one. If an AI answer points your customer to a fake store, you lose the sale, you may get the complaint, and trust in your brand takes the hit.

AI chatbots fake shopping sites study key numbers: 151 brands tested, 6,795 answers, 9 percent risky, 718 risky domains flagged

What the Podqi study tested

Podqi ran its tests during September 2026, according to Gadget Review and Startup Fortune. It asked the three chatbots shopping questions about 151 brands and collected 6,795 answers.

Startup Fortune reports that Podqi reviewed 4,437 domains that appeared in those answers and flagged 718 as risky: 351 fake sites and 367 dead domains. The coverage does not explain in detail how the prompts were written, and Podqi’s full report is not linked in the articles we read, so treat the exact rates as one company’s snapshot.

How each AI chatbot performed

The risky site rate varied a lot by platform. Perplexity had the highest rate at 15% of answers, ChatGPT came in at 8% and Gemini at 4%.

Chatbot Answers exposing a risky site
Perplexity 15%
ChatGPT 8%
Gemini 4%
All three combined About 9%

The bigger worry is how those sites were presented. Startup Fortune reports that risky sites were recommended 6.3 times as often as they were warned against, and in 57% of answers that included one, the chatbot actively suggested it for shopping. Gadget Review says ChatGPT flagged a risky site as possibly unreliable in only 24% of the cases where it surfaced one.

Two other patterns stand out. Questions about regional storefronts were the riskiest type at 16%, and smaller brands faced roughly double the risky link exposure of large brands. One example in the coverage: Perplexity named an incorrect page as the official Stussy site.

What Google, OpenAI and Perplexity said

None of the three companies disputed the core finding, according to Gadget Review. Google said Gemini shows source links so users can check information and acknowledged that AI systems can make mistakes. OpenAI said it is working to improve safeguards and reduce deceptive shopping results. Perplexity said it does not claim to be perfectly accurate and is focused on improving accuracy.

Limits of the fake shopping sites data

Keep the caveats in mind before you quote these numbers in a board deck. The study measured exposure to risky URLs, not confirmed fraud victims. It covers one month and 151 brands. And Podqi sells brand protection services, which Startup Fortune and Gadget Review both point out.

Still, the direction fits what marketers already see. AI answers now act as a front door for product research, and we have written about how ChatGPT chooses websites for its answers. If those systems can be fooled by a lookalike domain, brands need to watch them the same way they watch search results.

What brands should do now

Steps for brands to protect customers from fake shopping sites in AI chatbot answers

These are Brandella recommendations, not steps from the study or the AI companies.

  1. Test your own brand in AI answers. Ask ChatGPT, Gemini and Perplexity where to buy your products, including country specific questions such as “official store in Canada” or “buy in India”. Note every domain they mention.
  2. Check every domain you see. Is it yours, an authorised reseller, or unknown? Pay special attention to old domains you once owned, past campaign microsites and retired regional stores, because dead domains were half of the risky list.
  3. Renew or redirect old domains. Do not let former brand domains expire. Keep them registered and redirect them to the right live page.
  4. Make your official store obvious. Use the same official URL in your website, social profiles, marketplace listings and business profiles. Clear brand authority signals help AI systems and people pick the real site.
  5. Publish an official retailer list. A simple page listing your own regional stores and authorised resellers gives AI tools and shoppers a clear reference.
  6. Report fakes and keep records. Report impersonating sites to the hosting provider and domain registrar, and use each chatbot’s feedback option when it recommends a fake. Log what you found and when.
  7. Monitor monthly. Add AI answer checks to your routine brand monitoring, next to branded search. Our guide on how to increase AI search citations explains how to strengthen the signals that point to your real pages.

Why smaller brands should act first

The finding that smaller brands faced about double the risk makes sense. Big brands have years of links, coverage and reviews pointing to one official site, while a smaller brand may have fewer signals for an AI system to lean on. That gap is exactly where a lookalike domain can slip in.

The good news is that the fixes are cheap. A one hour audit of what AI chatbots say about where to buy your products, plus a list of old domains to renew, closes many of the easiest gaps before the holiday shopping season.

Santaji Gadepatil

Written by

Santaji Gadepatil

Founder of Brandella Journal

Santaji brings 25 years of hands on experience in SEO, Google Ads, PPC and digital marketing. He has helped businesses across industries build lasting organic visibility and run profitable paid campaigns, and shares that real world experience through Brandella Journal to help marketers and business owners grow.

  • SEO
  • Google Ads
  • PPC
  • Digital Marketing