A new open-source CAPTCHA system called Cerno aims to distinguish between human users and AI bots by analyzing mouse movement patterns rather than relying on visual puzzles. Developed as a TypeScript SDK, Cerno evaluates 12 behavioral features extracted from raw pointer events during maze navigation tasks.

Unlike traditional CAPTCHAs that frustrate users with distorted text or image selection, Cerno generates mazes using the Growing Tree algorithm and analyzes motor-control patterns like velocity, path efficiency, and pause counts. The system employs a proof-of-work SHA-256 hash prefix and ECDSA P-256 signatures for security, while maintaining trustless validation through server-side maze regeneration from PRNG seeds.

Cerno represents a significant shift in bot detection methodology by focusing on the nuanced differences between human and AI interaction patterns. The system's motor-control analysis examines features such as jerk standard deviation, angular velocity entropy, and timing coefficients of variation - metrics that prove challenging for current AI models to replicate convincingly. With its open-source architecture and React integration, Cerno offers developers a privacy-conscious alternative to traditional CAPTCHA systems that often rely on third-party services and user data harvesting.