Cyberbullying remains a significant issue in digital spaces, making its detection a critical concern. This research aims to identify cyberbullying based on factors such as religion, age, ethnicity, and gender using an English dataset sourced from Kaggle. To enhance the detection accuracy, ensemble model integrating four machine learning techniques Support Vector Machine (SVM), Random Forest, Adaboost, and Logistic Regression. Individually these models achieved accuracy of 76% for SVM, 78% for Random Forest, 76% for Adaboost, and 74% for Logistic Regression. However, combining them into an ensemble approach boosted the accuracy to 81%. The cross-validation tests further enhance the model’s consistency, maintaining the same accuracy level. These findings demonstrate the effectiveness of ensemble learning in detecting cyberbullying across various categories, contributing to safer online environment.