Before You Let AI Teach You About Sex, Here’s Why Its Mistakes Matter
/Introduction
“Is it normal if I don’t finish every time?” “Can someone tell if I’m faking it?” Whether or not you support AI, it can be a tempting tool when you’re searching for answers to questions too awkward to ask out loud. With apps like Snapchat implementing their own chatbots, AI is becoming an easily accessible source of sexual education for many young people wishing to maintain their anonymity (CITE). As teens and young adults turn to the comforts of AI in place of more reliable resources, it is crucial to identify its shortcomings early on. This will create space for collaboration alongside health professionals and sex educators to ensure the information it provides is accurate, responsible, and empathetic.
How Safe is It? The Benefits of Using AI for Sexual Education
On a surface-level, popular large language models (LLMs) tend to provide helpful information about clear consent cues, while also offering less consistent but still generally acceptable information on STIs (Vowels & Marcantonio, 2025; Yıldız & Söğütdelen, 2025). Those with experience using AI have reported that they value the chatbots’ lack of judgment, with some mentioning more willingness to disclose sensitive information, such as having condomless sex (Nadarzynski et al., 2021). Considering this accessibility and anonymity, who is most likely to rely on AI? Youth with immigrant, racialized, or low-income backgrounds have far less access to traditional sexual health education (Rotermann & McKay, 2024). For queer and disabled youth, typical “sex ed” is often irrelevant, and seeking out advice at home or online can be uniquely challenging (Davies et al., 2024; UBC Okanagan News, 2025). Ultimately, the greatest promise of AI in sexual education may be its ability to reach those who lack access to equitable resources.
Where Safety Falls Short: Concerns About AI in Sexual Education
While AI has the potential to be a useful sexual health resource, whether it can fulfill that duty is a separate question. Even after using an LLM designed for sexual health queries, participants expressed doubt that such tools can help those experiencing sexual health anxiety about aspects of health that are stigmatized (Nadarzynski et al., 2021). LLMs also struggle to navigate nuanced sexual situations, often ignoring nonverbal cues and failing to account for silent sexual compliance and sexual regret (Vowels & Marcantonio, 2025). Concerningly, many LLMs misinterpret consent scenarios involving intoxication, often describing them as shifting from non-consensual to consensual and offering different interpretations depending on the genders of the people involved. These patterns matter, as young minds may use these tools to frame their own sexual experiences and trauma. Snapchat’s built-in chatbot, for example, has repeatedly failed to provide crisis resources (Marcantonio et al., 2025). These issues highlight how AI is ill-equipped to handle high-risk sexual questions, although it will answer them anyway.
So, Where Do We Go From Here?
Whether you choose to use AI to answer sexual questions in your everyday life ultimately depends on whether you’re willing to risk receiving information that may be incomplete, incorrect, or emotionally insensitive. A significant portion of sexual and reproductive health professionals express concerns about the use of LLMs for sexual advice, reflecting the broader risk they pose to some of the youngest and most vulnerable people in our society (Nadarzynski et al., 2023). If these systems are truly becoming unavoidable sources of information, then it is essential that we ensure their accuracy and their ability to handle these topics sensitively alongside information and guidance from sexual health professionals and educators.
Jessica Anderson (she/her), BA (Honours) Psychology Specialization, Year 3, Queen's University
References
Davies, A., O’Leary, S., Prioletta, J., Shay, B., Bryan, M., & Neustifter, O. (2024). Sexuality education for disabled children and Youth in Ontario, Canada: Addressing epistemic injustice through school‐based sexuality education. Children & Society. https://doi.org/10.1111/chso.12843
Marcantonio, T. L., Avery, G., Thrash, A., & Leone, R. M. (2024). Large language models in an app: Conducting a qualitative synthetic data analysis of how Snapchat’s “my ai” responds to questions about sexual consent, sexual refusals, sexual assault, and sexting. The Journal of Sex Research, 62(9), 1905–1919. https://doi.org/10.1080/00224499.2024.2396457
Nadarzynski, T., Lunt, A., Knights, N., Bayley, J., & Llewellyn, C. (2023). “But can chatbots understand sex?” attitudes towards artificial intelligence chatbots amongst sexual and reproductive health professionals: An exploratory mixed-methods study. International Journal of STD & AIDS, 34(11), 809–816. https://doi.org/10.1177/09564624231180777
Nadarzynski, T., Puentes, V., Pawlak, I., Mendes, T., Montgomery, I., Bayley, J., & Ridge, D. (2021). Barriers and facilitators to engagement with Artificial Intelligence (ai)-based Chatbots for sexual and Reproductive Health Advice: A qualitative analysis. Sexual Health, 18(5), 385–393. https://doi.org/10.1071/sh21123
Rotermann, M., & McKay, A. (2024, January 17). Where do 15- to -17-year-olds in Canada get their sexual health information?. Statistics Canada. https://www.doi.org/10.25318/82-003-x202400100001-eng
Vowels, L. M., & Marcantonio, T. (2025). Exploring how large language models understand sexual communication in hypothetical sexual situations. The Journal of Sex Research, 1–23. https://doi.org/10.1080/00224499.2025.2547814
Wellborn, P. (2025, October 6). UBCO study finds sex ed fails 2SLGBTQIA+ students. UBC’s Okanagan News. https://news.ok.ubc.ca/2025/10/06/ubco-study-finds-sex-ed-fails-2slgbtqia-students/
Yıldız, H. A., & Söğütdelen, E. (2025). Ai Chatbots as sources of STD Information: A study on reliability and readability. Journal of Medical Systems, 49(1). https://doi.org/10.1007/s10916-025-02178-z