<p>Over 4.5 billion users worldwide experience algorithmically curated content, yet systematic evidence of psychological impacts remains fragmented. This creates urgent public health and policy challenges. To quantify AI-driven recommendation algorithms' effects on mental health through a two-study approach: (1) meta-analysis of published research (Study 1); (2) deep learning classification on an independent social media dataset (Study 2). Study 1 synthesized 30 studies (<i>N</i> = 47,892) examining anxiety, depression, loneliness, affective polarization, and self-esteem. Study 2 trained a CNN-LSTM model (87.3% accuracy) on 50,000 + independent social media posts from Reddit Mental Health Corpus and Twitter Academic API to identify vulnerability profiles. SHAP analysis provided model interpretability. Random-effects models revealed significant adverse effects: anxiety (<i>d</i> = 0.42), depression (<i>d</i> = 0.38), loneliness (<i>d</i> = 0.51, largest effect), affective polarization (<i>d</i> = 0.43), and self-esteem (<i>d</i> = -0.33). Adolescents on image-based platforms showed 57–71% larger effects. Deep learning identified three risk profiles, with high-risk users (19.6%) exhibiting clinically significant depression (PHQ-9 = 16.8). Passive consumption amplified loneliness (<i>d</i> = 0.52), while active engagement showed protective effects (<i>d</i> = -0.16). Algorithmic content curation is associated with meaningful psychological harms, particularly among vulnerable populations. Findings support evidence-based regulation prioritizing well-being over engagement maximization and demonstrate how AI methods can illuminate AI's own societal consequences.</p>