Conceptual Design of Intelligent Moderation in Media Platforms: A Comparative Study of YouTube, X, Aparat, and Divar

This article aims to propose a conceptual model for the intelligent moderation of fake news on media platforms, capable of explaining the relationship between artificial intelligence, human oversight, and user participation in the processes of identifying, evaluating, and reducing the dissemination of false and misleading content. The significance of this issue lies in the fact that social networks, despite their extensive role in information circulation and public participation, have also become environments for the rapid spread of misinformation, the polarization of public opinion, and the erosion of social trust. This study adopts a qualitative approach based on comparative analysis and documentary research. To this end, the content moderation policies, guidelines, and practices of four platforms—YouTube, X, Aparat, and Divar—were examined and coded. To complement the documentary data, semi-structured interviews were conducted with experts from the social media ecosystem, including product managers, technical, legal, and content specialists, as well as experts in the field of media platforms. The data were analyzed through three stages: open coding, conceptual aggregation, and categorization. The findings indicate that effective moderation of fake news on media platforms does not rely on a single-dimensional mechanism; rather, it operates as a multilayered and hybrid system. Within this system, artificial intelligence algorithms are used for the initial and scalable detection of content; human moderators are responsible for contextual assessment, identifying borderline cases, and making final decisions; users contribute to the monitoring process through reporting and feedback; and platform policies define the legal and operational framework for addressing violations. The comparative analysis also showed that domestic platforms, while drawing on international experiences, have incorporated local legal and cultural requirements into their moderation policies. The study concludes that intelligent moderation is most effective when a balance is established among the speed and scalability of artificial intelligence, the interpretive sensitivity of human oversight, the transparency of platform policies, and responsible user participation. Sole reliance on artificial intelligence increases the risk of contextual errors and algorithmic bias, while complete dependence on human moderation is constrained by limitations related to speed, cost, and psychological pressure. Therefore, the proposed model emphasizes a hybrid, accountable, context-sensitive, and reviewable approach to content moderation.