Cyberbullying has emerged as a prevalent concern due to the extensive usage of social media, subjecting users to detrimental interactions and online abuse. Cyberbullying represents a critical public health issue with profound implications for mental health, as it has been associated with an increased risk of various psychological and developmental disorders, including suicidal ideation. Despite its growing recognition, there is a lack of publicly available datasets and the methodologies employed to assess its occurrence. This lack of consensus hampers efforts to effectively address and mitigate the adverse effects of cyberbullying on individuals and communities. In this work, we developed an algorithm that scrapes data from the YouTube API and performs a series of binary classification models to determine the presence of cyberbullying comments. Our methodology integrates Natural Language Processing (NLP) approaches, including text normalization and sentiment analysis via Valence Aware Dictionary and sentiment Reasoner (VADER), with machine learning and deep learning models for binary classification. We performed sentiment analysis on the collected data using Support Vector Machine (SVM), Bi-directional Long Short-Term Memory (BiLSTM), Bi-directional Gated Recurrent Unit (BiGRU), and Bi-Directional Encoder Representations from Transformers (BERT) models. We also tuned the hyperparameters of these models for the best results. Initial findings indicate the efficacy of our proposed system, achieving 94% accuracy, and demonstrate its capacity to reduce the effects of cyberbullying through automated detection.