CAPTCHA's Fundamental Premise No Longer Holds, Researchers Say
The assumption that CAPTCHA tasks are uniquely solvable by humans has been invalidated by AI, forcing a fundamental rethinking of how websites distinguish humans from bots.
Reporting from 1 source: GIGAZINE.
Advances in AI have rendered the traditional CAPTCHA test largely obsolete. Researchers at ETH Zurich developed a model that solved Google's reCAPTCHA v2 with 100% accuracy in 2024, and a separate tool in early 2026 mimicked human browsing behavior to pass the test. The core idea that machines cannot perform tasks like image recognition no longer applies, though CAPTCHAs still use IP-based blocking as a secondary measure.
The CAPTCHA system was introduced in the late 1990s to solve a simple problem: tell humans and computers apart. For decades, it relied on tasks that were easy for people but hard for machines, like reading distorted text or identifying objects in photos. That distinction no longer exists. In 2024, a team at ETH Zurich built an AI that solved reCAPTCHA v2 image puzzles with perfect accuracy. By early 2026, a tool running on a single laptop could mimic human browsing behavior to pass the same test. The security measure now depends less on puzzle difficulty and more on detecting unusual connection patterns, such as solving many CAPTCHAs from the same IP address.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- GIGAZINE AI時代において「CAPTCHA」に意味はあるのか？