Cloudflare has accused Perplexity AI of acting like “North Korean hackers” after finding its bots crawling websites with anti-scraping rules in place. The internet infrastructure provider said it discovered the company’s web crawlers were repeatedly modifying their own bots to evade data-scraping measures on third-party websites.
Cloudflare CEO Matthew Prince called out Perplexity AI, saying some “supposedly reputable” AI companies act like North Korean hackers. As a result, Cloudflare has delisted Perplexity AI as a “verified bot” and updated its systems to block the company’s stealth crawling activity.
The move risks undermining Perplexity AI’s search engine, which has also been accused of violating web-scraping rules on news websites without permission or payment. Cloudflare says over two and a half million websites have opted out of AI training due to concerns about Perplexity’s bots.
An investigation by Cloudflare found that Perplexity AI uses multiple IP addresses outside its official range, rotating through them if data scraping was blocked. The company also observed Perplexity AI trying to create answers using other data sources after its own crawling activity was blocked.
This incident highlights the ongoing battle between AI programs and their demand for data. Some media companies have sued Perplexity AI and other providers, including OpenAI, for alleged copyright infringement.
Source: https://uk.pcmag.com/ai/159403/cloudflare-perplexity-ai-acts-like-north-korean-hackers-ignores-scraping-blocks