Social media cyberbullying is an issue that is getting worse and can lead to serious consequences for victims such as sadness anxiety and even suicide manual monitoring and reporting are the mainstays of conventional approaches for identifying cyberbullying which can be time-consuming and inefficient automating the identification of cyberbullying on social media is a possible answer. The purpose of this paper is to develop an effective machine learning-based approach for detecting online bullying on social media. The paper gathers a dataset of social media messages including instances of cyberattacks and analyze the data using NLP and machine learning. The results show that our method accurately and efficiently detects cyberbullying with an f1 score of 0.93 provide a full description of the most common types of cyberbullying and their frequency on social media platforms as well the findings point to the promise of machine learning-based techniques for locating and putting an end to cyberbullying on social media as well as for helping to create more effective and targeted intervention strategies.