Generative AI platforms are known for harvesting data from the web on a massive scale. Now they are becoming targets themselves. According to Decodo’s latest analysis, reported by TechRadar Pro (France) on 7 October 2026, ChatGPT and Perplexity rank among the five most scraped sites for the first time.
A largely reshuffled ranking
Seven of the ten sites in Decodo’s 2026 ranking are new entries. Only TikTok, in first place, Google, in second, and Amazon, in eighth, already featured in the previous edition. TikTok remains at the top with 18% of the observed activity, or almost one in five scraping operations.
ChatGPT enters straight in third place, with 14% of the observed activity. Perplexity takes fifth place, with 10% of the observed data. Conversely, several platforms that were in the 2025 top 10 drop out this year, including Walmart, eBay and Airbnb.
These figures come from Decodo and cover the activity the company observes; the TechRadar article does not detail the study’s methodology, nor does it link to it directly. It is therefore one player’s measurement, not an exhaustive census of scraping on the web.
Why collect AI answers
For Decodo, the shift reflects a change in the information organisations are looking for. Two years ago, explains its CEO, Vaidotas Juknys, companies scraped the web to see what users were searching for; now, he says, they scrape AI platforms to see what users are being told.
The source mentions several possible uses: comparing a product’s ranking with that of its competitors in ChatGPT’s answers, checking how a company is described there, measuring how often a brand is mentioned, or recording the sources cited by the assistant. In the background, consumers are turning to AI-assisted shopping journeys, while social networks remain a major avenue for observing consumer trends.
A legal question the study does not address
The TechRadar article does not comment on the lawfulness of such data collection, on the terms of use of the platforms concerned, or on any national law. For a Swiss company tempted by this form of monitoring, the question of the applicable framework nonetheless arises, whether it concerns the terms of use set by the providers of these services, data protection or competition law. The source does not settle it, and it reports no position from any authority: this is a point to be examined case by case, rather than a conclusion to be drawn from this ranking.
What Decodo expects for 2027
Decodo believes that Gemini and Copilot could in turn enter the top 10 in 2027. The company also envisages a deeper shift: AI agents that would trigger scraping operations themselves, rather than merely using the data collected. Next year, Vaidotas Juknys suggests, agents could be the ones making decisions instead of just reading the answers. This is a projection by the company, not an observation.
Sources
Adrien Bar Hiyé, Craig Hale, “ChatGPT et Perplexity deviennent pour la première fois certains des sites les plus scrapés du Web”, TechRadar Pro (France), 7 October 2026.
