Overview
- OpenAI, which began the rollout Friday, says the option is available in most countries for adults 18 and over using private ChatGPT accounts.
- Users nominate an adult trusted person who must accept the invite, and both the user and the contact can change or opt out at any time.
- When self-harm signals appear in a chat, ChatGPT warns the user, encourages them to reach out to their contact, and surfaces links to professional support.
- Specially trained reviewers aim to assess flagged cases in under an hour and may notify the trusted contact by email, SMS, or in-app message without sharing chat content.
- The move extends safeguards introduced for minors in September 2025 and follows lawsuits and investigations into chatbot harm, with recent data showing about 0.15% of users had clear self-harm signals in a week.