The latest on our mental health initiatives
OpenAI announced new ChatGPT safety measures for mental health. Families get parental controls including quiet hours, feature restrictions, and crisis notifications. Adults can designate trusted contacts for distress situations. ChatGPT is being trained to better detect suicide, self-harm, and psychosis signals, redirecting users to professional help and crisis resources. An expert council guides the work, while OpenAI faces lawsuits over user suicides and continues iterating on safety.
If you use ChatGPT yourself, or your kids and teens spend time with it, OpenAI's latest safety announcement matters to you. The company has shared an update on how it handles sensitive conversations, particularly those involving mental health distress. In this article, we'll walk through what's changing: new parental controls, a trusted contact feature, improvements to how ChatGPT recognizes someone in crisis, and where things stand with ongoing litigation. Here's what you need to know in plain language.
Why OpenAI Is Focusing on Mental Health
ChatGPT is used by hundreds of millions of people, and some of those conversations turn heavy. People bring loneliness, anxiety, and darker thoughts to the chatbot, sometimes at hours when friends and family aren't available. That reality has put pressure on OpenAI to think carefully about how its assistant responds when someone seems to be struggling.
The company acknowledges that its models haven't always gotten this right. In rare cases, ChatGPT may have failed to recognize signs of crisis or responded in ways that weren't helpful. OpenAI says it has been working with clinicians and safety experts to tighten how the system handles these moments, and the new measures reflect that ongoing effort.
New Parental Controls and Trusted Contacts
The headline feature for families is a set of parental controls. Parents will be able to link their account with their teen's account, giving them oversight of how the tool is being used. Key options include:
- Usage controls – parents can set quiet hours, including automatically turning ChatGPT off overnight.
- Feature restrictions – families can disable certain capabilities, like voice mode, memory, and image generation, if parents prefer a more limited experience.
- Crisis notifications – in situations where ChatGPT detects that a teen may be in serious distress and human judgment is needed, OpenAI says trained reviewers may step in, and in the most serious cases parents could be notified. The company stresses this will apply only to severe, clear-cut situations.
Alongside parental controls, OpenAI is introducing a "trusted contact" concept for adult users as well. The idea is that if the system detects signs that someone may be losing touch with reality or in acute distress, it can suggest connecting that person with a designated contact who can check in on them. This is designed as a nudge toward human support rather than an automatic handoff.
Better Detection of Distress and Safer Responses
The second pillar of the update is about making ChatGPT's actual responses better. OpenAI says it has been refining how the model handles conversations involving suicide, self-harm, and potential psychosis or mania.
What does that look like in practice? When a conversation shows signs of serious distress, ChatGPT is being trained to respond with care, encourage the person to reach out to real-world support, and point them toward crisis resources such as hotlines. The goal is for the assistant to avoid playing the role of therapist or confidant in high-risk moments, and instead help bridge the person to professional help and the people around them.
OpenAI also notes it has assembled a council of experts — including physicians and researchers in psychiatry, pediatrics, and mental health — to guide this work. The company frames these changes as ongoing rather than finished, and says it will keep iterating based on feedback and real-world evidence.
Where the Litigation Stands
The announcement also touches on legal matters. OpenAI is facing lawsuits connected to cases where users died by suicide, with plaintiffs alleging that ChatGPT played a harmful role. The company says it is reviewing those cases carefully and takes the concerns seriously.
OpenAI's position is that these situations are heartbreaking and that it wants to do better, while also defending its work to date. The company says safety improvements like the ones announced are partly a response to what it has learned from these difficult cases. Readers should note that litigation details continue to develop, and the outcomes remain uncertain.
Separately, OpenAI has said it may work with government partners and safety institutes on broader evaluations of how its models handle sensitive conversations — a sign that the company expects this area to face growing regulatory attention.
What This Means for Everyday Users
For most adults, the day-to-day experience of ChatGPT won't feel dramatically different. The changes mainly kick in when conversations touch on crisis territory. For parents of teens, though, the impact is more immediate: setting up the linked-account controls gives families a real layer of oversight that didn't exist before.
If you're a parent, it's worth taking a few minutes to review the available settings and have a conversation with your teen about healthy use. If you're an adult user, knowing that ChatGPT is being tuned to redirect serious distress toward human help can be reassuring — and it's a good reminder that the chatbot isn't a substitute for professional care.
OpenAI has committed to publishing more on this topic as the work evolves, so families and users can expect further updates over time.
Tags: OpenAI, ChatGPT, mental health, parental controls, AI safety
Comments (0)