AI sexuality education: an exploratory study of ChatGPT’s responses to adolescents’ authentic sexuality-related questions

Publicly available Large Language Models (LLMs) such as ChatGPT have undergone rapid adoption since their release in 2022. It is likely that a large number of adolescents turn to LLMs for sexuality education, especially after their integration into mainstream search engines. However, it remains unclear what information these models provide and which underlying logics shape their outputs. This paper examines how ChatGPT responds to adolescents’ questions about intimacy and sexuality. The study draws on 753 authentic questions collected during school workshops. From this pool, 34 prototypical questions were selected and entered into ChatGPT-4o. The responses were then analysed using documentary method, focusing both on the explicit information conveyed and the implicit norms reproduced. Findings indicate that ChatGPT explicitly aligns with the values of holistic and inclusive sexuality education, although with a strong emphasis on health and prevention. Implicitly, however, the model occasionally reinforces stereotypes, conservative norms, and restrictive sexual scripts, including both traditional heteronormative assumptions and emerging normative frameworks of sex-positivity and diversity. The paper discusses emerging double binds and reflects on their implications for sex and sexuality education, pedagogy, social work, and counselling.