This study presents an innovative approach to mitigate mobbing as a cyberattack by integrating artificial intelligence (AI) techniques such as Latent Dirichlet Allocation (LDA) and Robustly Optimized BERT Pretraining Approach (RoBERTa) according with the Cross-Industry Standard Process for Data Mining methodology (CRISP-DM). Mobbing in digital environments constitutes an emerging form of cyber attack that affects both individuals and organizations. Preventing mobbing requires a comprehensive approach involving legal frameworks, employee awareness, and technological solutions. This work uses AI to identify patterns of mobbing and proposes a structured process based on CRISP-DM to address this phenomenon. It focuses on analyzing mobbing through digital data to develop a content-filtering solution for platforms like email and messaging systems. The study identified eight general phases that describe the overall context of mobbing, as well as four specific phases that outline the detailed process of harassment within it, using fine-tuning techniques for detection. The results show how AI can automate the detection and mitigation of mobbing, minimizing its impact on victims and improving cybersecurity.