Agentic AI: New Scraping Threat & Unprepared Defenses

The barrier to large-scale web scraping has disappeared. Autonomous agents can now scrape websites with minimal configuration, making sophisticated attacks accessible to many. Here's the thing: previously, scraping required specialized skills like reverse engineering and proxy rotation. Now, these agents read a page, decide next steps, and complete multi-step tasks simply from a plain language goal. This turns scraping into a commodity available to anyone with a motive. What's interesting is that the same agentic AI tools used for productivity can also be used for hostile data extraction. Enterprise software applications using agentic AI are projected to jump from under 1% in 2024 to one-third by 2028. This means defenders are facing a broader range of attackers, equipped with commonly available tools. The bottom line: the demand for training data for large machine learning models has made the entire open web a target, threatening the business models of content-producing sites.