Abstract
<jats:p>In contemporary digital environments, social media platforms operate through artificial intelligence (AI)-driven systems, including algorithmic ranking, recommender infrastructures, targeted advertising architectures, and automated moderation processes. These systems do not merely organize content distribution; they also shape user attention, patterns of interaction, and exposure to digital risks. This study examines how AI-based platform ecosystems reconfigure opportunity structures for criminal conduct while simultaneously functioning as algorithmic gatekeeping mechanisms capable of mitigating harm. Drawing on routine activity theory, rational choice theory, and situational crime prevention, the article conceptualizes platforms as dynamic criminogenic environments that influence offending costs, the likelihood of detection, and victim vulnerability. The analysis identifies major forms of crime and harmful conduct associated with social media, particularly fraud, scams, harassment, cyberstalking, online abuse against children, deepfakes, and other forms of synthetic deception. It also explores central dilemmas of platform governance, including procedural safeguards, transparency, accountability, and the management of algorithmic bias, while reviewing key developments in European and international regulatory and liability frameworks. Special attention is given to the tension between platform efficiency and due process, especially where automated detection tools affect content visibility, user reporting, and account restrictions without adequate meaningful explanation or accessible avenues for review and redress. By linking investigative and evidentiary challenges to prevention-oriented governance mechanisms, the study formulates analytically grounded recommendations concerning safety by design, evidentiary readiness, and stronger child protection safeguards. It argues that effective responses to crime in AI-mediated platform environments require an integrated approach that combines criminological insight, technological governance, and rights-based regulation.</jats:p>