Even though online communication media serves as a platform for people to connect and discuss, it can also be misused to direct harmful and abusive comments that might affect an individual’s emotional and mental well-being. The proliferation of social media communication has made it incredibly impractical to manually weed out abusive content, necessitating the development of automated features to maintain greener and friendlier social forums. By leveraging text mining and a machine learning model that can accurately identify and label racist speech, we aim to achieve this goal. The implementation of such a system not only enhances user experience but also promotes a safer online environment. Continuous improvements and updates to the algorithm enable the model to adapt to new and evolving forms of harmful content, ensuring its sustained effectiveness. Our model has achieved an accuracy score of 94.94% and a precision score of 94.49% in classifying remarks as harmful or benign. This high level of accuracy and precision demonstrates the potential of machine learning in addressing the pervasive issue of online abuse, thereby contributing to healthier and more respectful online interactions.