ChatGPT's Trusted Contact: AI Alerts for Self-Harm Risk

ChatGPT is rolling out a new "Trusted Contact" feature to help users in mental health crises. If ChatGPT detects serious self-harm risk, it can now notify a pre-selected trusted person. This aims to add a human layer of support during difficult times. Here's how it works: Adult users can choose one contact, who must accept the invitation. If ChatGPT flags a conversation for self-harm, it first warns the user. Then, a trained human review team examines the chat. If they confirm a serious risk, the trusted contact gets an alert, but without chat details, only a message to check in. OpenAI aims to complete this review within one hour. This feature highlights the growing role AI plays in sensitive personal discussions.