Analyzing the availability of dangerous and destructive content in the main sources of information on the Internet for adolescents

Abstract

Introduction. Government policies on information security and the upbringing of schoolchildren in various countries include measures to restrict access to potentially harmful content on Internet resources. The relevance of this study lies in the need to implement these measures in educational practice by identifying destructive content and associated threats in social networks and other components of the digital society. Purpose of the study - analyze threats in social networks, assess the safety of Internet resources, and classify threats to develop practical recommendations for regulating interactions with potentially harmful Internet resources. Materials and Methods. The study employed approaches used in the development of modern social media monitoring systems (e.g., "Kribrum," "Vitok-OSINT"), publicly available social media APIs, text extractor classification, multidimensional, and cluster analysis to assess the accessibility of harmful and destructive information content for schoolchildren. Results. A methodology for threat assessment was proposed, based on the potential for physical harm, psychological damage, social contagion, prevention complexity, and long-term consequences. A taxonomy of threats was developed through clustering, identifying four categories: socio-destructive, communication-psychological, ideological-manipulative, and technical-criminal threats. A multidimensional analysis of harmful and destructive information revealed that such content is highly accessible on most popular non-gaming Internet resources frequented by schoolchildren. The clustering process identified risk groups with hazard ratings on a 10-point scale: socio-destructive (9.5/10), communication-psychological (8.7/10), ideological-manipulative (8.3/10), and technical-criminal (7.8/10). Conclusion. The study found that TikTok and Telegram pose the highest risks in terms of modern cyber threats, while VKontakte and RuTube are the least dangerous. The proposed taxonomy will enable the development of more advanced approaches to training neural networks for proactive filtering of destructive content. The recommendations can be used to enhance the qualifications of teachers and educators in ensuring the information security of schoolchildren amid the digital transformation of schools.