This study is a systematic review of artificial intelligence (AI) applications to support adolescent mental health. We searched six major databases and Google Scholar for studies published between January 2010 and July 2025, following PRISMA 2020 and a pre-specified internal protocol. Twenty-four studies met the inclusion criteria, covering CBT/psychoeducation-based chatbots; machine learning models for predicting depression, self-harm, and suicide attempts; natural language processing to detect symptoms from text; and passive sensing using mobile-phone and digital sensor data. Findings were synthesised using a narrative synthesis, and methodological quality was appraised with the Mixed Methods Appraisal Tool (MMAT). Overall, the available evidence suggests that AI-based interventions may be associated with small-to-moderate improvements in depressive symptoms, anxiety, and psychological distress, alongside potential gains in mental health literacy. However, only a limited proportion of studies used randomised controlled trial (RCT) designs, outcomes and methods were heterogeneous, and follow-up periods were generally short. Most studies were conducted in high-income countries, with relatively limited attention to Southeast Asian contexts, ethical governance, and equity of access. Taken together, AI shows promise for supporting adolescent mental health, but the certainty of evidence remains constrained; future research should prioritise longitudinal designs and more rigorous trials that are culturally responsive, equity-oriented, and embedded in robust ethical regulation.