A New Default: The Teen Experience
To address growing safety concerns, OpenAI has rolled out 'ChatGPT for Teens', a specialised experience for users aged 13 to 17. This isn't an optional mode; it's the automatic default for anyone who self-identifies as a teen during signup or is estimated
to be under 18 by the platform's systems. This approach aims to create a baseline of protection by default. The teen experience comes with stronger content restrictions, learning-focused features, and reminders to take breaks and use the tool responsibly. The goal is to guide teens toward healthier AI habits within an environment tailored to their developmental stage.
The Educated Guess: Age Prediction
Instead of just relying on users to state their age, OpenAI uses an age prediction model. This system doesn't ask for ID from everyone; rather, it infers a user's likely age by analysing various signals. These can include the topics they discuss, the time of day they are active, and other general usage patterns. If the model estimates a user is likely under 18, it automatically moves them into the more protected teen experience. OpenAI has stated that when its system is uncertain about a user's age, it will err on the side of caution and apply the stricter teen safeguards. Adults who feel they have been incorrectly categorised can go through a verification process to restore full access.
Content Filters and Stricter Guardrails
At the core of ChatGPT's safety measures are its content filters, which are significantly enhanced for the teen experience. This 'Reduce sensitive content' setting is on by default for teens and is designed to limit exposure to a range of inappropriate topics. These include graphic violence, dangerous viral challenges, and content promoting unhealthy dieting or body shaming. Furthermore, the model is specifically trained to refuse to engage in romantic or sexualised roleplay and to avoid language that suggests it has feelings or consciousness, a measure meant to prevent unhealthy emotional dependence.
Parental Controls and Safety Alerts
For parents seeking more direct involvement, OpenAI has introduced optional parental controls. By linking their account with their teen's, a parent or guardian can manage certain settings. These controls do not allow parents to read their child's conversations, preserving a degree of privacy. However, they can set 'Quiet Hours' to limit access to the app and 'Study Hours' to encourage the use of learning-focused features. A critical feature is the safety notification system. If the platform's systems and human reviewers detect a conversation that indicates a serious risk, such as self-harm, a notification may be sent to the linked parent. This provides an alert in high-risk situations, giving parents information needed to support their teen's safety.
Limitations and the Human Element
No automated system is perfect, and OpenAI acknowledges the limitations. The effectiveness of these tools relies on continuous improvement and learning from real-world use. Determined users may still find ways to circumvent filters, and age prediction is an estimate, not a certainty. This is why the system includes more than just automated filters. Features like 'responsible homework reminders' are designed to redirect teens toward learning tools like Study Mode instead of just providing answers. Furthermore, the entire safety ecosystem relies on both automated monitoring and trained human reviewers who can intervene in the most critical cases, such as when threats of violence or serious self-harm are detected.
