In the rapidly evolving realm of digital communication, the escalating prevalence of toxic comments poses a formidable challenge, encompassing offensive language that targets individuals or groups based on various attributes such as race, gender, religion, or beliefs. This research underscores the urgent and critical need for advanced tools capable of autonomously detecting and mitigating toxic comments across diverse digital platforms. Beyond serving as protective shields for individuals, these tools emerge as invaluable allies for content moderators, significantly enhancing their efficiency in managing user-generated content and contributing substantively to the ongoing battle against online toxicity. The study meticulously explores an array of methodologies, spanning traditional machine learning approaches to cutting-edge deep learning models like BERT and XLNet. Notably, XLNet emerges as a standout performer, showcasing exceptional capabilities and innovative self-attention mechanisms that enable the nuanced capture of intricate patterns within toxic comments, thereby contributing significantly to the success of this complex task. However, the research extends beyond the confines of academic exploration; it endeavors to make a tangible impact by fostering a safer, more inclusive online environment that benefits all users. Through a commitment to continuous research and innovation, the aim is to address the pervasive issue of toxic comments and uphold the integrity and safety of digital communication spaces. With XLNet as a powerful and efficient ally, significant strides are being made towards achieving this crucial and overarching objective, representing a progressive stance in the ongoing quest to create a digital landscape free from the harmful effects of toxic online interactions.