OpenAI just rolled out a new safety feature called 'Trusted Contact' for ChatGPT. The idea is straightforward: you can designate someone who gets notified if the system picks up on conversations that suggest you might be at risk of self-harm.
This is OpenAI expanding beyond just showing crisis hotline numbers when sensitive topics come up. They're building an actual intervention mechanism into the product, which is a bigger step than most AI companies have taken so far.
For anyone using ChatGPT regularly, especially as a thinking partner or journal-like tool, this matters because these conversations can get personal. The line between processing difficult thoughts and needing real help isn't always clear, and AI companies are realizing they need systems for that gray area.
The feature is opt-in, so you have to actively set it up and choose your contact. That's important because it means OpenAI isn't making these calls unilaterally, but it also means the feature only works if people actually use it.
This fits into a larger pattern of AI companies wrestling with the fact that their tools are becoming genuine daily companions for millions of people. When that happens, product features start looking more like social infrastructure, and the stakes get higher fast.
