OpenAI introduces a trusted contact safety feature in ChatGPT

OpenAI has started rolling out Trusted Contact, an optional safety feature in ChatGPT designed to help connect adult users with real-world support during moments of serious emotional distress.

The feature allows users to nominate one trusted adult, such as a friend, family member or caregiver, who may receive a notification if OpenAI’s automated systems and trained reviewers detect that the user may have discussed self-harm in a way that indicates a serious safety concern.

OpenAI said the feature is intended to add another layer of support alongside existing safeguards in ChatGPT, including prompts that encourage users to contact crisis hotlines, emergency services, mental health professionals, or trusted people when appropriate. The company stressed that Trusted Contact does not replace professional care or crisis services.

Users can add a trusted contact through ChatGPT settings. The contact receives an invitation explaining the role and must accept it within one week before the feature becomes active. Users can later edit or remove their trusted contact, while the trusted contact can also remove themselves.

If ChatGPT detects a possible serious self-harm concern, the user is informed that their trusted contact may be notified and is encouraged to reach out directly. A small team of specially trained reviewers then assesses the situation before any notification is sent.

OpenAI said notifications are intentionally limited and do not include chat details or transcripts. Instead, they share the general reason that self-harm came up in a potentially concerning way and encourage the trusted contact to check in. The company said every notification undergoes human review and aims to review safety notifications in under one hour.

The feature was developed with guidance from clinicians, researchers and organisations specialising in mental health and suicide prevention, including the American Psychological Association. OpenAI said Trusted Contact forms part of broader efforts to improve how AI systems respond to people experiencing distress and connect them with real-world care, relationships and resources.

Why does it matter?

Trusted Contact points to a broader shift in AI safety away from content moderation alone toward real-world support mechanisms for users in moments of vulnerability. As conversational AI systems become part of everyday personal reflection and emotional support, companies face growing pressure to define when and how they should intervene, how much privacy to preserve, and what role human review should play in high-risk situations.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!