ChatGPT Self-Harm Alerts: New Trusted Contact Feature

ChatGPT is introducing a new "Trusted Contact" feature to help users at risk of self-harm. This allows adult users to register one person, like a family member or friend, in advance. If ChatGPT detects serious self-harm risk signals, it can help connect the user to real-world support through this contact. The system will first inform the user and offer direct outreach options. A specially trained reviewer will then assess the situation. If a real safety concern is confirmed, a notification will be sent to the trusted contact via email, text, or app. Importantly, the original conversation content will not be shared. Users can modify or delete their trusted contact at any time. This feature aims to connect users experiencing difficult moments with real-world support.