Recently, organisations have been focusing just on external attacks, neglecting the threat posed by insiders. Insider threats are now one of the most massive and damaging threats that could happen for any system. Also, the log files are the target for any insider attack. Therefor, This project proposes a secure machine learning system for log insiders attacks to determine the normal and abnormal behaviours after scanning the log files, and then send warning when detecting any irregular action. So, the paper introduces a Web application that highlights the idea of securing systems from internal malicious doubts using some machine learning algorithms like(SVM-Isolated forest-Kmeans). Those algorithms briefly collects raw logs, parsing those logs, then extract features and finally detect any danger with applying risk assessment techniques to minimize risks to be with high effectiveness in evaluation matrix stage. At last, the impact that the paper aims is to focus on the idea of implementing an applicable and essential detector application to stop precariousness.

•  Constant Monitoring for log files to detect insiders unexpected behaviors.

•  Determine the difference between normal and abnormal behaviors using machine learning and to secure cooperate ’s system from the abnormal ones.

•  Get an early warning before system getting attacked.

•  Secure the system from the malicious insiders using the application.

•  The system keeps constant monitoring of system’s logs.

•  Detection of any abnormal behavior happening by an insider.

•  Measuring the effectiveness of machine learning models.

•  The system sends a warning to stop any threat on the system.

•  Securing large corporations by using Anomaly Detection application.