AuthenShield: CAPTCHA-Free Bot Detection Through Passive User Interaction Analysis

Abstract

The importance of cybersecurity has increased significantly in the last decade, and there is a growing demand for safe modes of online identity verification. Traditional text-based and audio-visual CAPTCHAs are becoming vulnerable to AI-based hackers. They also hamper the user's browsing experience and pose accessibility challenges for disabled people. To mitigate these obstacles, this study presents a thorough analysis of passive verification methods employing behavioral biometrics to differentiate between humans and bots. A review of existing literature is conducted to evaluate the effectiveness of mechanisms such as keystroke dynamics, mouse movement tracking, and eye-blink recognition through facial detection. Keystroke analysis is carried out using a dataset of timing metrics from multiple users. Facial detection integrates real-time tracking of eye movements and blinking by computing the Eye Aspect Ratio (EAR). These data points are processed through machine learning models (such as SVM, XGBoost, and Random Forest Classifier) run using Python to train the classifiers and achieve an enhanced, multilayered bot-detection system. Honeypot traps are also incorporated to enhance bot resistance. A DDoS attack is simulated to evaluate the system's efficiency and resilience, and the results indicate that AuthenShield offers superior protection against automated attacks while continuing to maintain accessibility and consumer experience.