Today, cloud computing has become an important technological paradigm that has become widely implemented in the activities of modern organizations, including in Ukraine. The transition to cloud services (CS) allows companies to increase efficiency, flexibility and competitiveness, as well as to optimize operational costs and risks related to information security (IS). Cloud technologies ensure the availability, scalability and reliability of corporate software applications and systems, which contributes to their widespread use in the modern business environment. At the same time, the use of CS creates new challenges and threats to IS, among which internal violators occupy a prominent place. Internal IS threats can pose the greatest danger to companies. These can be both intentional actions by disloyal employees and unintentional mistakes by honest employees. Insider attackers can have direct access to sensitive information and systems, making it difficult to detect and counter these threats. Effective internal threat risk management requires a comprehensive approach that includes technical, organizational, and personnel security measures. This work is an attempt to develop a method for early detection of such violators based on the application of Bayesian networks. The classification of internal intruders proposed in the study and the identified indicators of their activity constitute a new approach to the analysis and detection of IS threats in the cloud environment. The implementation of the learning model in Python and GeNIe Modeler demonstrates the possibility of creating effective intrusion detection tools that can complement existing DLP systems. The use of modern development and modeling tools makes this work relevant and innovative in the field of protecting cloud services from internal information security threats. Further research involves detailing the proposed method, as well as the analysis of other mathematical approaches that can be used to solve the task, with an assessment of the results of their application.