OpenAI just announced Trusted Contact, an optional feature that lets ChatGPT users assign an emergency contact who gets alerted if concerning conversations come up. We're talking about discussions around self-harm or suicide that the AI flags as potential crisis situations.
The feature is only available to adult users, and you can designate friends, family, or caregivers as your contact person. OpenAI says the approach is built on expert guidance that connecting someone in crisis with a trusted person they already know can actually help.
This sits alongside the localized crisis helplines that ChatGPT already surfaces when these topics come up. So it's not replacing those resources, just adding another potential support layer.
For anyone using AI assistants regularly, especially in personal or reflective contexts, this is OpenAI acknowledging that these tools are becoming intimate enough to surface real mental health concerns. The company is trying to build guardrails that go beyond just showing a phone number.
The opt-in nature matters here. You have to actively set this up, which means it won't surprise anyone with unexpected notifications. But it does raise questions about how AI companies balance user privacy with duty of care as these tools get more deeply embedded in daily life.
