Websites face a growing crisis from AI scraper bots, overwhelming servers with traffic as they harvest data for large language models. These malicious actors employ millions of unique IP addresses, often disguised as human users, making them difficult to block. The traffic predominantly originates from compromised residential and mobile networks, a practice known as "residential proxies."

Operators of these proxy networks range from criminal enterprises, like the recently dismantled IPIDEA botnet, to companies offering "ethically sourced" IPs. Prominent among the latter is Bright Data, which advertises its ability to bypass access controls and traffic limits, often through "free" VPN services that route user traffic through their network. Media-streaming devices are also emerging as a significant source of compromised endpoints for these operations.

While major AI companies often conduct their own, more transparent scraping, the vast majority of overwhelming traffic stems from unknown entities funding these residential proxy attacks. The methods to combat this include proof-of-work systems like Anubis, commercial services requiring human verification, and data poisoning techniques. However, these defenses impose a significant burden on both site operators and legitimate users, creating a constant arms race to protect the open web.