ChatGPT adds Trusted Contact alerts for serious self-harm risk cases

Trusted Contact

OpenAI is rolling out an optional ChatGPT safety feature that lets adults nominate one trusted person who can be notified if reviewers determine a conversation indicates a serious self-harm risk.

# ChatGPT adds Trusted Contact alerts for serious self-harm risk cases

## Opening summary

OpenAI is starting to roll out Trusted Contact, an optional ChatGPT setting that allows an adult user to nominate one person they trust to receive a warning if a conversation suggests a serious self-harm risk. The company says the feature is layered on top of its existing crisis support prompts and still routes every notification through trained human review before anything is sent.

## Main article

According to OpenAI’s announcement, users over 18 can choose a single adult contact in settings, and that person must accept the invitation before the feature becomes active. If OpenAI’s systems detect a potentially acute self-harm situation, ChatGPT warns the user that it may notify their trusted contact and encourages them to reach out directly first.

The escalation is not fully automated. OpenAI says a trained review team checks the situation before an email, text message, or in-app alert is sent, and that the notification is intentionally limited to the general concern instead of exposing the user’s conversation transcript.

Engadget adds the broader context that OpenAI has been under pressure to improve how ChatGPT responds to people in distress. That makes Trusted Contact notable not because it turns ChatGPT into a crisis service, but because it adds a deliberate bridge from the chatbot to an actual person in the user’s life.

## Why it matters

This matters because AI safety features are more credible when they help users reconnect with real-world support instead of pretending the model alone is enough. Trusted Contact shows OpenAI trying to build a narrower and more accountable response path for one of the hardest categories of chatbot use.

## Source notes

- Verified against OpenAI’s official announcement, including age rules, invitation flow, human review, and the no-transcript privacy limit. - Verified against Engadget, which matches the feature description and adds context about OpenAI’s recent safety scrutiny. - The article avoids medical or legal claims and does not frame the feature as a replacement for professional help.