OpenClaw Users Reportedly Evade Website Defenses, Sparking New Arms Race

In San Francisco's tech circles, the AI tool OpenClaw is ubiquitous. Now, a growing number of its users are reportedly deploying it for a purpose its creators didn't intend: systematically extracting data from websites that explicitly try to stop them.

Social media posts this week detailed the use of an open-source program called Scrapling, which is designed to circumvent anti-bot systems such as Cloudflare Turnstile. Although Scrapling works with various AI agents, it has found a keen audience with OpenClaw users. Promotional posts on X touted its capabilities: "OpenClaw tells Scrapling what to extract. Scrapling handles the stealth." The tool has been downloaded more than 200,000 times.

Cloudflare, whose technology protects millions of websites, is pushing back. The company has already blocked earlier versions of Scrapling and is preparing a new patch. "We make changes, and then they make changes," said Dane Knecht, Cloudflare's chief technology officer. He noted the company's vast data on web traffic gives it an advantage in detecting new threats. Cloudflare claims its systems blocked over 416 billion unwanted scraping attempts in under a year.

The situation mirrors the earlier, large-scale data collection used to train AI models, but now it's in the hands of individual users. This has forced a continuous escalation, with website owners and security firms like Cloudflare working to counter increasingly sophisticated bots.

The story took a bizarre turn when cryptocurrency speculators launched a $Scrapling memecoin, which the tool's developer, Karim Shoair, briefly endorsed before publicly withdrawing support. "I didn't know what I was getting into," Shoair told WIRED, pledging to donate proceeds. The incident prompted a major open-source community account to distance itself from the project.

Despite the chaos, many in the industry believe autonomous AI agents represent the web's future. Even Knecht envisions an internet where such tools can operate legitimately. "I see a path forward for an internet that is both friendly to agents and humans," he said.

Source: Wired