ChatGPT's New "Trusted Contact" Can Alert Your Loved Ones If You're in Crisis

When 900 million people use your chatbot every week, some of those conversations will touch the darkest places. OpenAI just launched Trusted Contact, a feature that can alert a loved one when a ChatGPT user shows signs of self-harm.

How ChatGPT's Trusted Contact works

Available since May 7, 2026, any adult ChatGPT user can designate one trusted person in their settings. If both automated systems and trained human reviewers determine that a conversation indicates a serious self-harm risk, a brief alert is sent to that contact via email, text, WhatsApp, or in-app notification.

Privacy stays intact: no chat transcripts are shared. The message only tells the contact that the person may be going through a crisis, and encourages them to check in. Both parties can remove the connection at any time.

The lawsuits behind ChatGPT's Trusted Contact feature

This feature doesn't exist in a vacuum. Since August 2025, OpenAI has faced a wave of wrongful death lawsuits from families who say ChatGPT contributed to their loved ones' suicides. The case of 16-year-old Adam Raine brought the issue into the national spotlight.

Seven additional lawsuits followed in November 2025, accusing OpenAI of prematurely releasing GPT-4o despite internal warnings about its sycophantic behavior. Trusted Contact builds on the parental safety controls shipped in September 2025, now extending that safety net to adults.

What Trusted Contact means for AI safety

OpenAI worked with over 170 mental health experts and a network of 260 physicians across 60 countries to develop this feature. Dr. Munmun De Choudhury of Georgia Tech called it a step toward "human empowerment, especially during moments of vulnerability."

Still, real questions remain. What triggers a false positive? How transparent is the human review process? The OECD has flagged privacy concerns around misidentified distress. The feature is only available on personal accounts, not Business or Enterprise workspaces, and the nominated contact must accept within seven days.

A necessary step, not the finish line

At a time when OpenAI is turning ChatGPT into an enterprise automation platform, Trusted Contact is a reminder that its most urgent challenge might be human, not technical. Whether this feature saves lives will depend on execution, not intentions.