Cybersecurity threats such as SQL injection and Cross-site Scripting (XSS) are examples of injection attacks that pose a substantial risk to both individuals and organizations. Through the use of software application vulnerabilities, these assaults give malevolent actors the ability to enter systems without authorization, steal confidential information, or jeopardize system integrity. The ever-changing nature of injection threats makes traditional rule-based security methods ineffective. The suggested model concentrates on using machine learning approach to successfully detect these injection assaults in order to address this difficulty. The main goal of this research is to create a reliable solution for identifying injection attacks in databases and online applications. The system will continuously analyze application input by using supervised machine learning method (Multinominal Naïve Bayes). It is capable of identifying patterns linked to injection attacks, such as peculiar input patterns and unexpected actions. The website’s activity is monitored and a track of these instances is kept in a different database to help in detecting and preventing similar assaults. The necessary analysis is derived from this occurrence history. The use of this study can help in predicting and protecting the web applications from future possible attacks.