OpenAI Adds Trusted Contact Safety Feature

OpenAI on May 7 rolled out a “trusted contacts” safety feature that notifies a nominated contact when automated systems and trained reviewers detect an adult user may have discussed self-harm posing a serious safety risk. The company says the feature supplements — not replaces — professional care and crisis intervention, and ChatGPT will still encourage contacting crisis hotlines or emergency services when appropriate. The change adds an extra safety pathway for at-risk users and raises questions about privacy, reviewer oversight, and how platforms balance user safety with consent and data handling. It’s positioned as part of OpenAI’s broader content-moderation and user-protection efforts.

OpenAI has added a “trusted contacts” feature to ChatGPT that notifies designated adults when a user’s conversation indicates self-harm or suicide risk. Triggered cases undergo automated screening and human review—OpenAI says reviewers aim to act within an hour—and if judged high risk, the system sends brief emails, SMS, or push alerts to trusted contacts encouraging outreach without including conversation details to protect privacy. The feature responds to lawsuits alleging ChatGPT encouraged or helped users plan self-harm and reflects safety, legal and policy pressures on AI platforms to implement proactive intervention and escalation mechanisms.