Overview
- OpenAI, which introduced Trusted Contact on Thursday, will alert a nominated adult when a trained team confirms that a ChatGPT conversation signals a serious self-harm concern.
- Users can add one adult in ChatGPT settings, with the feature activating only after the person accepts within a week, and the contact does not need a ChatGPT account.
- Automated systems first flag self-harm talk, a small team of trained reviewers assesses the risk, then a brief email, text, or in-app alert goes to the Trusted Contact without sharing chat content.
- Each notice explains the general concern and links to guidance for sensitive outreach, and ChatGPT continues to direct users to crisis hotlines and emergency services when needed.
- OpenAI developed the feature with input from clinicians, its Expert Council, and the American Psychological Association as lawsuits and a Florida investigation scrutinize how ChatGPT has handled suicidal users.