Recent estimates from NPR based on data from OpenAI indicate that over a million weekly users engage in discussions about suicide and self-harm using ChatGPT. Recognizing the importance of human connection in suicide prevention, OpenAI introduced a feature called Trusted Contact in May. This optional feature aims to connect users experiencing distress with someone they trust.
NPR reported on this feature following a tragic story of a young woman who died by suicide after using ChatGPT. OpenAI spokesperson Gaby Raila explained that ChatGPT is trained to recognize potential signs of a user’s risk by identifying indirect cues. When a significant safety risk is detected, the app notifies the registered Trusted Contact, assuming the user has set it up. Despite claiming good adoption, OpenAI did not provide specific figures on usage.
NPR sought to understand how users become aware of this feature. A producer, Megan Lim, tested the app but found no initial mention of Trusted Contact. Only after directly inquiring did she receive confirmation about the feature’s existence. She learned it is opt-in and must be configured in Settings. If not set up, the bot cannot notify anyone, but still encourages users to seek support through loved ones or hotlines, like 988.
Mental health experts express mixed feelings about this feature. Holly Wilcox from Johns Hopkins suggests offering it to all users before emergencies occur. Ursula Whiteside, of Now Matters Now, found the feature simple to use but noted that distressed users might struggle with setup. Whiteside’s research shows that individuals benefit from having a trusted contact option when planning ahead.
Jennifer Lockman, director of a suicide prevention clinic, points out potential challenges. Individuals contemplating suicide often feel they are burdens. A chatbot might not overcome these feelings or replace professional guidance in selecting contacts. Yet, Lockman agrees that anything fostering real human connection is beneficial, emphasizing the protective nature of such bonds.
OpenAI continues refining when and how ChatGPT introduces the Trusted Contact feature in cooperation with psychologists and researchers. Along with this feature, OpenAI has added safeguards such as recommending breaks and providing crisis hotline information. While Lockman hopes the feature benefits users, she stresses the importance of research to evaluate its effectiveness in saving lives.
If you or someone you know is in crisis, contact the Suicide & Crisis Lifeline at 988.
Leave a Reply