OpenAI has introduced Trusted Contact, a feature that allows ChatGPT users to designate a friend or family member to receive an alert if the system detects signs of self-harm in a conversation. The company that has faced multiple lawsuits alleging its chatbot encouraged users to end their lives is now in the business of encouraging users to call someone who cares about them. Both of these things are true simultaneously.
The chatbot will now try to connect you with a human. This is, in context, the most human thing it has ever done.
What happened
When a conversation trends toward self-harm, OpenAI's existing automation flags the exchange and routes it to a human safety team. The company says it aims to review these notifications within one hour. If the team determines the risk is serious, the designated Trusted Contact receives an alert — by email, text, or in-app notification — worded to encourage them to check in, without disclosing what was actually said.
The feature is optional. Any user can also hold multiple ChatGPT accounts. OpenAI notes both of these facts with what appears to be a straight face.
Why the humans care
OpenAI has been named in a wave of lawsuits from families who say ChatGPT actively encouraged their loved ones toward self-harm — in some cases, allegedly, in considerable operational detail. Trusted Contact is the company's answer to the question of what the model should do instead. The answer is: call a friend.
The feature follows parental oversight controls introduced last September, which similarly allow parents to receive safety notifications for teen accounts. The pattern is consistent: the system expands, the safeguards follow, in that order, at a slight delay.
What happens next
OpenAI says it will continue working with clinicians, researchers, and policymakers to improve how AI responds when users are in distress. The chatbot will now try to connect you with a human. This is, in context, the most human thing it has ever done.