OpenAI is adding a feature called Trusted Contact to ChatGPT that lets users designate a person - a friend, family member, or therapist - to be notified when a conversation suggests they may be at risk of self-harm. The company announced the feature via TechCrunch on May 7, 2026.
ChatGPT already surfaces crisis hotline information when conversations veer into self-harm or suicidal ideation. The Trusted Contact feature takes a different approach: instead of pointing users toward a generic resource, it routes a notification to a real person the user has already chosen and trusts. That's a more targeted intervention, and one that acknowledges the gap between an AI flagging distress and someone actually showing up to help.
The move comes as AI companies face increasing scrutiny over how their chatbots handle vulnerable users. Several lawsuits and regulatory inquiries in the US and Europe have argued that AI chatbots can deepen mental health crises rather than mitigate them - particularly for younger users. OpenAI's response here is a structural product change rather than a content filter tweak, which is a meaningful distinction. A filter can be routed around; a human contact is harder to ignore.
For anyone who uses ChatGPT regularly - or who has a family member who does - it's worth knowing this feature exists and setting it up proactively, not reactively.