The observed increase in the number of threats in the field of information security of society and the state creates the need for their timely identification. The paper proposes a combined approach to identifying and categorizing agents that have the greatest impact on users in social networks, which consists in combining and consistently applying existing methods. A simulation model of information dissemination was created and tested, which made it possible to simulate the process of information interaction between participants of the Internet information and telecommunications network and the influence of malicious agents on them. The results of the experiment show that changing the configuration of the agency network of influence and increasing the psychological protection of users to destructive effects allows you to reduce the scale of distribution of negative content