AI-BASED FAKE LOGIN ATTEMPT DETECTION SYSTEM USING BEHAVIORAL ANALYTICS

Abstract

In the age of the increasingly sophisticated cyber threats, the conventional login security mechanisms are often inadequate to combat with the surge of the fake login attempts using valid credentials. This happens with credential leaks, phishing or brute-force automation bypass the traditional security filters that poses a serious risk to organizations and the user data. In this paper we will introduce a machine learning-based fake login detection system that operates by evaluating behavioral login patterns, including the parameters such as login time, location, device type and the access frequency. The solution we propose includes employing a Random Forest classification algorithm that is trained on a custom dataset of the 1,000 simulated login sessions which is generated for the 50 virtual users across varied geographic and the temporal conditions. The proposed system successfully identifies the suspicious login attempts with the accuracy rate of 94.3% that offers a reliable second layer of the authentication beyond passwords or tokens. This research underscores the growing relevance of the behavioral analytics in the enhancement of the authentication systems of enterprise settings, where unauthorized access may lead to the sensitive data breaches. The results of the research validates that the integration of AI with the traditional security frameworks strengthens defense dynamically in real-time environments