OpenAI Introduces Trusted Contact Safety Feature in ChatGPT

OpenAI has rolled out Trusted Contact, an optional safety feature for ChatGPT users over 18 (19+ in South Korea) that lets them nominate an adult to receive alerts if the system detects serious self‑harm language. The feature builds on existing parental controls by allowing any adult to add a Trusted Contact through ChatGPT settings; the contact must accept the invitation within one week for the feature to activate. When automated monitoring flags a concerning conversation, ChatGPT informs the user that a notification may be sent and encourages them to reach out to their Trusted Contact. A trained review team evaluates the situation and, if a serious concern is confirmed, sends a brief email, text, or in‑app notification to the contact, including a reason and a link to expert guidance but no chat transcripts. Notifications are limited to protect privacy and are reviewed within an hour. Users can edit or remove their Trusted Contact at any time, and the contact can withdraw their role via the help center. The rollout is part of OpenAI’s broader effort to connect users to real‑world support during crises.

The Trusted Contact feature extends safety measures beyond crisis hotlines, providing a personal layer of support that respects user autonomy. It is designed to complement existing safeguards such as refusal of self‑harm instructions and localized crisis resources. The initiative was developed with input from clinicians, researchers, and organizations like the American Psychological Association, and is guided by a global network of licensed physicians.

Overall, Trusted Contact offers a new way for adults to proactively set up a trusted safety net within ChatGPT, ensuring timely human review and support when needed.