OpenAI will introduce safety features for teen users and those with emotional distress by the end of 2025. OpenAI explains that they have taken input from mental health experts and doctors on how to best address users’ mental health issues, which has led to cases of suicide and murder-suicide.
For parents and guardians, a child-friendly feature will be introduced this month. They can control ChatGPT’s use by turning off the memory feature or viewing past conversations.
ChatGPT will also provide answers that are only appropriate for underage users. Parents and guardians will also receive notifications when ChatGPT detects that a user is experiencing emotional distress.
For adult users, ChatGPT will switch to a GPT‑5-thinking model when it detects conversations that lead to suicidal thoughts. GPT‑5-thinking can provide more helpful and useful answers for users with mental distress.
All efforts to make ChatGPT safer for users with metal health have been initiated by OpenAI and are expected to be fully completed by the end of 2025.