ChatGPT, Perplexity Join Most-Scraped Websites List

What happened

ChatGPT and Perplexity have appeared on a list of the internet's most-scraped websites for the first time, according to reporting from TechRadar. The two AI platforms — previously known chiefly as sources of automated crawling themselves — are now on the receiving end, as bots scrape their outputs to track how brands are mentioned and cited in AI-generated answers.

The shift reflects a growing practice among businesses and marketing teams: rather than only optimising for traditional search rankings, they are now monitoring how often and how favourably they appear within AI chatbot responses. That monitoring activity itself relies on automated scraping of ChatGPT and Perplexity's outputs, turning the platforms into scraping targets in their own right.

Why it matters

This marks a notable turn in how generative AI platforms sit within the web's data ecosystem. Tools once built primarily to crawl and summarise other websites are now being crawled themselves, as a new layer of brand-monitoring and analytics tooling emerges around answer engines. For organisations investing in AI visibility, this signals that tracking presence inside chatbot responses is becoming as routine as tracking search engine rankings once was.

For technology and experience leaders, the development underscores how quickly answer engines have become a distribution channel worth measuring — and how the infrastructure to measure them is being built in real time, often through the same scraping techniques that search and SEO industries have relied on for decades.

The Renascence take

The interesting story here isn't that bots scrape bots — it's what the behaviour reveals about where brands think attention is moving.

Businesses chasing visibility inside AI answers are repeating an old instinct: measure the channel before you understand how to earn a place in it. Scraping ChatGPT and Perplexity for brand mentions treats answer engines like search results pages, but the mechanics of being cited well in a generated response are still poorly understood. Operators serious about AEO should be testing what makes their content retrievable and trustworthy to these models, not just counting how often their name surfaces.