OpenAI has started introducing a new ChatGPT feature known as Trusted Contact. The tool gives adult users the option to choose someone close to them — such as a family member, friend, or caregiver — who can receive an alert if the platform identifies conversations related to self-harm that may point to a serious risk. The system works through a combination of automated monitoring tools and human safety reviewers designed to detect potentially concerning interactions.
Users can select one adult to serve as their Trusted Contact through the ChatGPT settings menu. Once added, the chosen person receives a request explaining what the feature involves and must approve it within seven days before it becomes active. If the request is rejected or expires, the user can invite someone else instead.
To be eligible, the Trusted Contact must be at least 18 years old in most countries, while users in South Korea must choose someone aged 19 or above.
How does it work?
When ChatGPT’s safety systems detect conversations that may indicate a serious risk, the platform first alerts the user that their Trusted Contact could be informed. It also encourages the user to connect with someone directly by offering suggested ways to start the conversation.
After that, a dedicated group of trained safety reviewers examines the flagged interaction. If they determine that the situation involves a genuine safety concern, the selected Trusted Contact is sent a short alert through email, text message, or an in-app notification for users who have a ChatGPT account.
The alert sent to a Trusted Contact does not contain copies of conversations:
The alert sent to a Trusted Contact does not contain copies of conversations or any detailed chat history. Instead, it briefly explains why the notification was triggered and provides access to professional resources designed to help people handle sensitive situations and conversations. OpenAI says the review process and notification system are designed to work within roughly an hour.
Users also have full control over the feature and can change or remove their Trusted Contact whenever they want through ChatGPT settings. The selected contact can also opt out of the feature through the support centre.
Trusted Contact builds on the parental safety tools already available for teen accounts linked to parents or guardians. OpenAI says the feature was created with guidance from clinicians, researchers, and organisations such as the American Psychological Association. The company also worked with its Global Physicians Network, which includes more than 260 licensed doctors across 60 countries.
The new feature joins several existing safety measures already available in ChatGPT, including links to crisis support services, restrictions on harmful self-harm-related content, and responses shaped with input from mental health professionals. Trusted Contact is currently being rolled out for users who are 18 years of age or older.