The use of online learning platforms has experienced significant growth since the COVID-19 pandemic, resulting in an increased workload for teachers and an increase in toxic behaviors in virtual environments. To address these challenges, an AI-based automated moderation system is proposed, which employs the re-trained Toxicity model from the TensorFlow library. This tool allows real-time detection of offensive comments, threats and attacks, adjusting the sensitivity of moderation according to the educational context and offering support for multiple languages. This implementation aims to create a safer and more respectful learning environment, alleviating the moderation burden faced by teachers and enabling more positive interactions between students.