Adaptive Behavioral Authentication for Bot Detection Using ML

Automatic bots increasingly compromise the security and reliability of online platforms, requiring effective detection mechanisms. This paper proposes a behavioral-based framework to distinguish human users from the bots through the analysis of keystroke dynamics and mouse movement patterns. A dedicated web-based interface was developed to capture real-time interaction data, including writing speeds and cursor trajectories. These features were collected in a structured dataset and used to train machine learning for accurate prediction. Experimental results indicate a high classification accuracy, which shows the framework's capability to identify non-human interactions. The integration of behavioral metrics with machine learning contributes to robust, adaptive bot detection, and enhancing the integrity and security of the digital systems.