<![CDATA[This study analyzes the YouTube recommendation algorithm as a new social agent that influences user dependency in Medan City. The research focuses on the effects of YouTube usage intensity, content consumption patterns, algorithmic awareness, and digital habitus on YouTube dependency, with digital literacy positioned as a mediating variable. This study employed an explanatory quantitative approach with a cross-sectional survey design. The sample consisted of 400 YouTube users aged 18-45 years in Medan City selected through proportionate stratified random sampling. Data were collected using a structured questionnaire and analyzed through Structural Equation Modeling-Partial Least Squares (SEM-PLS). The findings show that usage intensity, content consumption patterns, algorithmic awareness, digital habitus, and digital literacy have positive and significant effects on YouTube dependency. Usage intensity and content consumption patterns also significantly affect digital literacy, whereas algorithmic awareness and digital habitus do not directly affect digital literacy. All indirect paths through digital literacy are significant, indicating that digital literacy functions as an ambivalent mediating mechanism: it improves digital competence while also strengthening attachment to the platform. These findings confirm that the YouTube algorithm should not be understood merely as a technical feature, but as a sociotechnical actor that shapes users' digital habitus, choices, and dependency.]]>
