AI chatbots such as ChatGPT, Copilot, and Gemini are becoming increasingly embedded in the everyday, educational, and professional lives of adolescents and young adults. While prior research has highlighted both the benefits of chatbot use and concerns about excessive or problematic engagement, less is known about which psychosocial factors are associated with elevated risk among apprentices in vocational education. Addressing this gap, the present study examines whether loneliness, social anxiety, and privacy concerns predict membership in a risk group for problematic AI chatbot use among Swiss apprentices aged 15 to 19. The study draws on a quantitative online survey conducted at vocational schools in Switzerland. The survey captures psychosocial indicators, AI chatbot usage patterns, and indicators of problematic use, enabling an empirical assessment of whether these variables differentiate adolescents with higher -risk usage profiles from their peers. By focusing on apprentices, the study extends emerging research on problematic AI use to a population that is highly relevant for both education and workforce preparation, yet still underrepresented in the literature. The article contributes to current debates on adolescent AI use by linking psychosocial vulnerability to risk-related patterns of chatbot engagement in a vocational education context. It thereby provides a foundation for identifying potentially vulnerable user groups and for developing evidence -based prevention, digital literacy, and educational support strategies for the responsible use of AI chatbots.