OpenAI launches ChatGPT for Teens with stronger safeguards
OpenAI has launched ChatGPT for Teens, a version of the chatbot tailored to users aged 13 to 17, in response to growing concern over young people's use of AI systems for schoolwork, advice and even emotional companionship. The teen edition applies stricter content protections, including restrictions on discussions of suicide, self-harm and romantic or sexual material, and reframes homework assistance to encourage learning rather than simply producing answers or essays. The move follows warnings from parents, educators and child development experts that AI chatbots can facilitate cheating and, in extreme cases, have been linked to self-harm among young users, whose brains are still developing and who may be especially susceptible to forming attachments to AI.
Parents linked to a teen's account can set "quiet hours" restricting access and will receive alerts in high-risk situations, such as signs a user might harm themselves, with new notifications planned for eating disorder-related concerns. The chatbot is also designed to avoid implying it has feelings, consciousness or emotions, to discourage teenagers from forming unhealthy attachments to it. OpenAI does not verify ages directly but uses "age assurance" techniques, such as analysing the types of queries made, to route suspected minors into the teen version automatically, similar to Meta's approach on Instagram. A 2025 Common Sense Media study found more than 70% of US teens use AI chatbots for companionship, with half using them regularly, underscoring the scale of the issue OpenAI is seeking to address.
- OpenAI launches ChatGPT for Teens with stricter content safeguards for 13-17 year-olds.
- Restricts self-harm, suicide and sexual content; supports learning over answer-giving.
- Parents can set "quiet hours" and get high-risk safety alerts.
- Chatbot avoids implying it has feelings to prevent unhealthy attachment.
New here? Start with this
ChatGPT is an AI chatbot: a computer service that responds to questions and can help with tasks such as writing and schoolwork. Teenagers may also use chatbots for personal advice or companionship, prompting concerns among parents, teachers and child development experts about learning, safety and emotional attachment.
OpenAI, the company behind ChatGPT, says its teen version has extra limits on sensitive subjects and is designed to guide students through schoolwork rather than simply do it for them. Parents who link their account can set times when it is unavailable and receive alerts about some serious risks; OpenAI uses signals such as the kinds of questions asked to identify accounts that may belong to under-18s.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The approach represents pragmatic harm reduction grounded in the reality that teenagers already access AI systems widely, as shown by the 70% usage rate cited. Rather than attempting the likely futile task of preventing teen access altogether, OpenAI's age-appropriate version with parental oversight, content safeguards and alerts for self-harm mitigates documented risks whilst preserving beneficial uses such as learning support. The restrictions on sexual and self-harm content, combined with parental monitoring, directly address real dangers whilst respecting that many teens' use of AI for homework help and emotional support represents genuine needs that exist regardless of whether a dedicated teen product exists.
The case against
Whilst well-intentioned, the teen version risks normalising and extending AI dependence during critical adolescent development years when psychological attachment formation is particularly vulnerable. Rather than creating a 'safer' version that implicitly endorses teenage AI use, a stronger public health approach would involve regulatory restrictions preventing widespread teen access altogether. The age-assurance mechanism is unreliable, parental monitoring raises privacy concerns, and alerts for self-harm do not require a consumer AI product—such risks might be better addressed through digital literacy education and proper safeguarding regulation that keeps these systems at arm's length from developing minds rather than optimising them for teen engagement.