ChatGPT Introduces Trusted Contact for Enhanced Security

A New Feature to Enhance User Security

OpenAI has recently introduced a new feature in ChatGPT called Trusted Contact. This optional security option allows adult users to designate a trusted person, whether a friend, family member, or caregiver. The goal is to notify this person if the system detects concerning discussions related to self-harm. By integrating this feature, OpenAI aims to provide additional support to users in crisis by facilitating a connection with a trusted individual.

This initiative builds on previous safety notifications aimed at parents or guardians of teenagers. Now, anyone over the age of 18 can choose to add a Trusted Contact to receive alerts in case of signs of distress. This expansion of security options aims to strengthen the support network around adult users.

How Trusted Contact Works

According to experts, social connection is a crucial factor in reducing the risk of suicide. Trusted Contact is designed to encourage users to connect with someone they already trust. While it does not replace professional care, it serves as an additional layer of protection. ChatGPT continues to recommend that users contact helplines or emergency services if necessary.

Users can add an adult as a Trusted Contact through the ChatGPT settings. In South Korea, users must be at least 19 years old to use this feature. The designated person will receive an invitation explaining their role and must accept it within a week. If they decline, the user can choose another person.

When the system detects potentially concerning discussions about self-harm, ChatGPT informs the user of the option to notify their Trusted Contact and encourages them to reach out. A specialized team then reviews the situation. If they believe there is a serious risk, a notification is sent to the Trusted Contact via email, text message, or through the app if they have a ChatGPT account.

The notification is intentionally brief to protect the user's privacy. It does not include chat details or transcripts but indicates the general reason for the alert and encourages the Trusted Contact to check in. Expert guides are also provided to help manage these sensitive conversations.

Users can modify or remove their Trusted Contact at any time, and the contact can also withdraw through the help center. Although these situations are rare, the system is designed for a quick response. Each notification undergoes a trained human review before being sent, and OpenAI strives to review these safety notifications in under an hour.

Safety at the Heart of Every Interaction

In addition to Trusted Contact, ChatGPT incorporates several safety measures to manage sensitive conversations. The system encourages users to contact emergency services or mental health professionals as needed.

OpenAI has collaborated with over 170 mental health experts to improve the detection of signs of distress and guide users toward appropriate support. ChatGPT can also suggest taking a break after prolonged use to promote healthy tech habits.

The system is trained to refuse harmful requests, such as instructions for self-harm, and redirects users to local crisis resources.

Towards Continuous Improvement of Safety

Trusted Contact is part of OpenAI's efforts to develop AI systems that provide real help in difficult times. The company continues to collaborate with clinicians, researchers, and policymakers to enhance AI systems' responses to user distress. The goal is to ensure that AI does not operate in isolation but helps connect users to essential care and resources.