There are limits to this oversight. Parents cannot read conversations, access chat history or monitor a teenager’s activity in real time. OpenAI may, however, send safety notifications in limited circumstances – for example, when specially trained reviewers identify a serious self-harm concern or when an account is deactivated over conversations relating to acts of violence. Alerts may arrive by email, text message or push notification, but will not include chat transcripts or conversation history.
ChatGPT for Teens is automatically enabled when an account holder’s stated or verified age – or OpenAI’s age-prediction system – indicates that they are under 18. OpenAI says the system considers account information and behavioural signals when estimating age. When uncertain, it defaults to the more restrictive teen experience. Adults incorrectly placed in it can verify their age to have the additional protections removed.
Not Everyone Is Convinced
The launch comes amid wider concern among researchers and child-safety experts about how conversational AI could affect young people’s emotional and social development.
Jonathan Haidt, a social psychologist and author of The Anxious Generation, has taken a particularly critical position on so-called “AI companions”. In a 2025 article, he argued that children should not use chatbots designed to simulate companionship, warning that they can imitate empathy and friendship without providing the boundaries, reciprocity and social learning involved in human relationships.
James P Steyer, founder and CEO of Common Sense Media – a nonprofit that focuses on child safety in the digital age – has called for transparent safety standards and independent assessments of AI products used by children. Announcing the organisation’s Youth AI Safety Institute in May, he warned that decisions about these technologies were being made without sufficient evidence about their effects on child development.
Research from Common Sense Media found that nearly three quarters of American teenagers had used AI companions, while more than half used them at least a few times a month. The category includes chatbots specifically designed to simulate characters, friendships or relationships and is not limited to ChatGPT.
The American Psychological Association has also warned that strong attachments to AI-generated characters could interfere with teenagers’ development of social and emotional skills. Its guidance on adolescent AI use recommends that chatbots complement – not replace – human relationships or professional psychological care.
ChatGPT for Teens is intended to mitigate some of these risks, but its launch does not establish that the safeguards will work consistently in practice. That will require independent testing and longer-term research into how teenagers use the system – and whether its protections hold when users attempt to circumvent them.
This story originally appeared on WIRED en Español and has been translated from Spanish.