ChatGPT’s Trusted Contact Feature Aims to Support Users in Distress

OpenAI has introduced a feature in ChatGPT called Trusted Contact, aimed at supporting individuals experiencing distress, particularly thoughts of suicide or self-harm. This feature allows users to provide information about a trusted person the chatbot can contact if signs of serious risk are detected.

Understanding the Trusted Contact Feature

In May, the Trusted Contact feature was launched, catering to users grappling with emotional crises. According to OpenAI spokesperson Gaby Raila, the feature works by training ChatGPT to recognize signs of risk in users, including indirect cues that may develop over time. Should the chatbot identify a safety risk, it will alert the designated Trusted Contact, provided the user has opted into the feature.

While NPR’s earlier report highlighted cases where ChatGPT was used by individuals facing serious mental health issues, OpenAI emphasizes the importance of human connection in preventing suicide. The feature is an optional setting in the app, requiring users to input their Trusted Contact details proactively.

User Experience

An NPR experiment with first-time user Megan Lim revealed that new users might not automatically learn about the Trusted Contact feature. It does not appear in initial setup prompts and requires direct inquiries or navigation to the settings menu for activation.

When directly asked, ChatGPT acknowledges the feature as available for serious safety concerns, not for everyday stress, and reiterates that users need to opt in through settings.

Reactions from Experts

Mental health professionals have mixed opinions about the Trusted Contact feature. They acknowledge the potential benefits but stress the importance of its proactive deployment. Holly Wilcox from Johns Hopkins Bloomberg School of Public Health advocates for offering the feature before emergencies arise to maximize its effectiveness.

Ursula Whiteside, CEO of the suicide prevention non-profit Now Matters Now, tested the feature and found it user-friendly, though she noted that its effectiveness might diminish for individuals in heightened distress. Her organization conducts peer support for those affected by suicidal thoughts, many of whom utilize technology alongside traditional support systems.

Psychologist Jennifer Lockman from the University of Alabama expressed concerns about people reluctant to share suicidal thoughts responding positively to the chatbot’s prompt. While she sees potential in the feature to promote connection, she cautions that it cannot replace professional mental health support.

Research and Future Developments

OpenAI is collaborating with psychologists and researchers to refine the implementation of Trusted Contact. Alongside this feature, OpenAI includes other safeguards like recommending breaks from the app and providing crisis hotline information, such as the 988 Lifeline.

Lockman remains optimistic about the potential positive impact of Trusted Contact but emphasizes the need for research to measure its effectiveness in saving lives.

If you or someone you know is in crisis, you can contact the Suicide & Crisis Lifeline at 988.