OpenAI has announced new measures to enhance safety for teen users on ChatGPT, prioritizing protection over privacy and freedom in certain cases.
The company detailed these updates in a blog post on Tuesday, highlighting steps to make ChatGPT safer for minors while continuing to improve overall security and usability for all users.
Balancing Privacy, Freedom, and Teen Safety
OpenAI CEO Sam Altman emphasized that some company principles can conflict. In these cases, teen safety takes precedence. Altman explained:
-
Privacy: ChatGPT conversations are increasingly personal, similar to speaking with a doctor or lawyer. OpenAI is implementing advanced security to protect user data, even from employees, while exceptions exist for serious misuse, threats to life, or potential societal-scale harm.
-
Freedom: Adult users are afforded broad usage flexibility. For example, ChatGPT can support adults in creative writing about sensitive topics or respond to requests that involve flirtatious dialogue, within safe limits. OpenAI aims to treat adult users “like adults” while preventing harm.
-
Teen Protection: Safety of minors is prioritized. Users under 18 (ChatGPT is intended for ages 13+) will experience content restrictions, including blocks on flirtatious talk and discussions of self-harm, even in creative contexts. In cases of imminent risk, parents or authorities may be contacted.
Altman noted,
We realize that these principles are in conflict and not everyone will agree with how we are resolving that conflict. These are difficult decisions, but after talking with experts, this is what we think is best and want to be transparent in our intentions.
Key Tools for Teen Safety
1. Age Prediction
- OpenAI is developing a system to estimate whether a user is over or under 18.
- Teen experiences are tailored to be age-appropriate, blocking graphic sexual content and providing additional protection in cases of acute distress.
- When the system is unsure of a user’s age, it defaults to the under-18 experience. Adults can verify their age to unlock full access.
2. Parental Controls
- Parents can link accounts with teens (minimum age 13) using an email invitation.
- Parents can guide ChatGPT responses, manage features like memory and chat history, and receive notifications if the teen experiences acute distress.
- If a parent cannot be reached in rare emergencies, law enforcement may be contacted.
- Blackout hours can be set to prevent teen use during specific times.
- In-app reminders for all users to encourage breaks during long sessions.
Continuing Commitment
OpenAI emphasized that these updates are part of an ongoing process, guided by feedback from experts, advocates, and policymakers. The company stated: “Your voices are shaping this work in ways that will matter for hundreds of millions of people.”
These measures aim to provide a safer ChatGPT experience for teens while maintaining privacy and freedom principles for adult users.