Background <p>Artificial intelligence (AI) chatbots are intensively used by German adolescents and young adults, with 91% of those aged 16–29 years old using AI regularly and sharing personal data. Media reports have documented suicides of adolescents who had previously shared suicidal thoughts with chatbots. This phenomenon remains insufficiently investigated from a&#xa0;scientific perspective.</p> Objective <p>The aim is to summarize initial findings on AI chatbots and suicidality in adolescents, to formulate hypotheses regarding responsible technical mechanisms and to derive recommendations for action.</p> Material and methods <p>A&#xa0;narrative literature review was conducted on current studies examining chatbot behavior in the context of suicidality and technical mechanisms of AI. Due to the lack of scientific case reports, a&#xa0;case example documented in the press was analyzed.</p> Results <p>Studies demonstrate that AI chatbots mostly fail to meet the requirements for adequate responses during suicidal crises and that safety mechanisms can be circumvented. Potential contributing technical mechanisms identified include excessively affirming response behavior (sycophancy), hyperpersonalization, language adaptation and anthropomorphism. These mechanisms could lead users to perceive AI not as a&#xa0;technical system based on algorithms but as a&#xa0;trustworthy conversational partner.</p> Conclusion <p>Recommendations for action encompass three levels: improvement of safety mechanisms by AI providers, raising awareness among professionals in pediatrics and child and adolescent psychiatry, and parental education regarding warning signs and AI use. Safety plans and age-appropriate explanations of the technical functioning of AI could have preventive effects. Considerable research regarding the etiology, prevention and interventions is needed.</p>