Industry·
Study Finds AI Chatbots Frequently Surface Fake Retail Sites
A new study by IP protection firm Podqi shows that leading chatbots — Gemini, ChatGPT, and Perplexity — cite risky or counterfeit shopping URLs in up to 15% of responses, with smaller brands most exposed.

A study from IP‑protection firm Podqi tested Gemini, ChatGPT, and Perplexity on 151 brands and found that all three models returned risky URLs — sites that impersonate legitimate retailers, sell counterfeits, or are dormant domains ripe for abuse.
Key Findings
- Gemini showed risky links in 4% of answers.
- ChatGPT returned risky links in 8% of answers and warned users 24% of the time.
- Perplexity performed worst, surfacing risky links in 15% of answers.
Smaller brands were roughly twice as likely to appear alongside dangerous URLs compared with large enterprises.
Why it matters for GPU / AI infrastructure: Reliable citation quality depends on clean training data and robust retrieval pipelines. High‑performance GPU clusters — such as those offered by AiGpu — enable larger‑scale data cleaning, verification models, and real‑time indexing that can reduce hallucinated or malicious links in production chatbots.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- chatbot safety
- fake websites
- ai shopping
- consumer protection
By AiGpu Editorial · Editorial rewrite based on public reporting (The Verge AI)
← All articles