In an increasingly digital world, ensuring secure and user-friendly authentication methods is paramount. Traditional password-based authentication systems are susceptible to security breaches, as passwords can be easily stolen or guessed. This paper explores the use of keystroke dynamics—unique typing patterns—as a secondary, non-blocking authentication layer to distinguish between genuine and fake users who enter the correct password. Unlike conventional systems that reject access based solely on incorrect passwords, our approach detects anomalies in typing behaviour to trigger additional verification steps, such as one-time passwords (OTPs). By analysing features such as inter-key delays, dwell times, and overall typing rhythm, we developed a machine-learning model capable of creating and updating a baseline typing profile for each user. Experimental results show that our system effectively identifies deviations in typing patterns with high accuracy, reducing the risk of unauthorised access without compromising user experience. This research demonstrates that keystroke dynamics can enhance security by leveraging behavioural biometrics, offering a promising direction for future authentication technologies.