ChatGPT can reach out to a friend if you're at risk of self-harm

Published: (May 8, 2026 at 08:17 AM EDT)
2 min read
Source: Engadget

Source: Engadget

Screenshots showing ChatGPT's new Trusted Contact feature.

Overview

OpenAI has introduced Trusted Contact for ChatGPT, allowing users to nominate a friend that the company can contact if they are at risk of harming themselves. The feature builds on ChatGPT’s parental controls and is available to adults (18+).

Background

  • Many people use ChatGPT as a digital therapist, with more than a million of its 800 million weekly users expressing suicidal thoughts in conversations, according to the BBC.
  • In 2023, OpenAI faced a wrongful‑death lawsuit alleging that the chatbot had helped a teenager plan a suicide.
  • A BBC investigation published in November 2025 reported at least one instance where ChatGPT advised a user on how to kill herself. OpenAI subsequently said it had improved how the model responds to people in distress.

How Trusted Contact Works

  1. Nominate a contact – Users can add one adult as their Trusted Contact in the ChatGPT settings.
  2. Invitation – The nominated contact receives an invitation and must accept it within one week; otherwise, the user can choose another contact.
  3. Warning – Before any notification is sent, ChatGPT warns the user that OpenAI may notify their contact if a serious risk of self‑harm is detected. The system also suggests conversation starters to encourage the user to reach out directly.

Human Review

  • The process is not fully automated. A “small team of specially trained people” reviews each case.
  • Only if the team determines there is a serious risk does OpenAI send the contact an email, text message, or in‑app notification.

Notification Content

“[The user] may be going through a difficult time. As their Trusted Contact, we encourage you to check in with them.”

The contact can view details about the warning, such as the detection of a suicide‑related conversation, but no transcript is shared to protect user privacy. OpenAI states that every notification undergoes human review and aims to complete the review within one hour.

Resources

If you or someone you know is experiencing suicidal thoughts, contact the National Suicide Prevention Lifeline at 1‑800‑273‑8255. The line is open 24/7, and online chat is also available.

0 views
Back to Blog

Related posts

Read more »

Running Codex safely at OpenAI

Overview As AI systems become more capable, they increasingly act on behalf of users. Coding agents can now: - Autonomously review repositories - Run commands...