Seeking Help, Facing Harm: Auditing TikTok's Mental Health Recommendations

Collection of recent arXiv outputs with several findings relevant to platform safety, manipulation, and analysis tools. A 7-day audit of TikTok shows engagement-driven amplification of mental-health content (~45% daily recommendations) and persistent low-but-nonzero exposure to potentially harmful suicide/self-harm content, indicating recommender insensitivity to help-seeking intent and a need for context-aware safeguards. A study on ResearchGate documents coordinated citation-farming via equal-reference groups that can materially inflate citation counts, threatening research-metric integrity.