A Systematic Literature Review of Generative AI Risks and Harms for Youth

As generative AI gains traction among youth, many associated risks—and strategies to mitigate them—remain poorly understood and are often studied in isolation. Through a systematic review of 95 research articles, we synthesize key trends and gaps in the youth—generative AI risk literature. We find that harms related to education and misinformation are well documented. However, prior work rarely examines how youth may use generative AI in ways that cause harm to others, or how risks unfold over time and within relationships. In particular, relational dynamics such as cyberbullying and mental health–related harms remain underexplored. To address this gap, we develop a sociotechnical–ecological model that links risk types to the contexts that shape them, and we discuss five implications for youth–centered generative AI governance and design. Together, these contributions lay a foundation for future research and interventions that support safer, developmentally appropriate generative AI use.