OpenAI’s new ChatGPT feature can alert a trusted contact during mental health crises

OpenAI has debuted a new Trusted Contact feature for ChatGPT, allowing users to designate an adult contact the company can reach out to if it detects a risk to their user’s mental health in conversation. Trusted Contact is rolling out to ChatGPT users from May 7, and is available to users globally. 

"Expert guidance⁠ identifies social connection as one of the most important protective factors to reduce suicide risk. Trusted Contact⁠ is designed to encourage connection with someone the user already trusts. It does not replace professional care or crisis services, and is one of several layers of safeguards to support people in distress. ChatGPT will still encourage users to contact crisis hotlines or emergency services when appropriate," OpenAI's safety team explained in a blog post.

Trusted Contact is limited to one adult over the age of 18 (or 19 in South Korea) who can be selected from ChatGPT's settings. It is not necessary that the trusted contact have a ChatGPT account, though it does make things easier when it comes to sending a notification. Once the trusted contact has been agreed, if a user is having some serious discussion of self harm, ChatGPT can prompt the user in question to reach out to their trusted contact, or proactively reach out to their contact themselves after a small team has reviewed the situation.     

AI companies are running into the limits of emotional dependency

The launch comes as AI companies face growing scrutiny over how chatbots interact with vulnerable users. In recent years, systems like ChatGPT and Claude have repeatedly been criticized for validating delusional thinking, reinforcing paranoia, or emotionally escalating fragile users rather than grounding them.

The issue has become prominent enough that terms like “AI psychosis” have entered wider public discussion, reflecting growing concern over how emotionally persuasive conversational AI can become when users treat it as a confidant, therapist, or companion.

OpenAI appears acutely aware of that pressure. In its announcement, the company specifically acknowledged that some users have experienced severe mental health crises while interacting with AI systems.

"While these serious safety situations are rare, when they do arise, our systems are designed to support timely review and response. While no system is perfect, and a notification to a Trusted Contact may not always reflect exactly what someone is experiencing, every notification undergoes trained human review before it is sent, and we strive to review these safety notifications in under one hour," OpenAI's safety team wrote.

There’s an argument to be made here that the increased nannying of users by Ai companies sensitive to lawsuits may backfire as those who are at risk the most find workarounds to avoid tripping the censors. But it’s clear that Trusted Contact, or something like it, was going to be developed anyway. There's a balance to be struck between babying users and letting their every thought be validated by AI, sure, but it's not clear where the line is – and we’re going to be stumbling along in the dark till we find it.