Overview

Trusted contact is an optional safety feature in ChatGPT that allows someone who trusts you to name you as a person they would want contacted if they may need additional support.

This feature is used only for safety concerns. If automated systems and trained reviewers suggest the person may be talking about suicide in a way that signals a serious safety concern, ChatGPT may notify you and encourage you to check in with them.

Availability

To participate, you must be 18 or older* and accept the invitation within 1 week. You do not need a ChatGPT account to be someone’s trusted contact.

What it means to be a trusted contact

Your role is to check in, listen, and help the person connect with additional support if needed.

Small actions can help. Reaching out, expressing care, and helping someone access more support can make a meaningful difference.

You are not expected to act as a counselor, crisis responder, or sole source of help. You are one part of a broader support network that may include friends, family, mental health professionals, helplines, and emergency services.

Accept a trusted contact invitation

You may receive a trusted contact invitation by email, text message, WhatsApp, and/or in the ChatGPT app.

To accept the invitation:

1. Open the invitation.

2. Review the explanation of the role.

3. Select Accept.

If you do not want to participate, see: Declining or removing yourself as a trusted contact.

Understand what notifications mean

After you accept the invitation, ChatGPT may later send you a brief notification. These notifications may be sent if automated monitoring systems and trained human reviewers determine that the person may be talking about suicide in a way that indicates a serious safety concern.

These notifications are meant to encourage a check-in. You are not solely responsible for the person's safety. If you are concerned about someone's immediate safety or want additional support, call local emergency services or a crisis helpline.

Notifications are intentionally limited to protect privacy, and a person’s ChatGPT conversations will not be shared. No system is perfect, and a notification may not always reflect exactly what someone is experiencing.

Know what ChatGPT shares and does not share with you

When you are invited, the invitation includes the person's name and email address so you can reach out to them.

If ChatGPT later sends a notification about a serious safety concern, the notification lets you know that the person may have discussed suicide with ChatGPT in a way that indicates a serious safety concern.

These notifications are intentionally limited to protect privacy. ChatGPT does not share the person's ChatGPT conversations with you.

Responding to a notification

If you receive a trusted contact notification, the goal is to check in with care, listen, and help the person connect with more support if needed.

When you are worried about someone's mental health, it can be hard to reach out, especially if you are not sure what to say. You do not need the perfect words.

A calm, direct, and caring check-in is usually the best place to start.

Start the conversation

Try to find a private moment to talk together. Lead with care, then share what you have noticed.

If you are reaching out after a notification, you can keep the conversation general to start.

Examples include:

• "I have been thinking about you. Would now be a good time to talk?"

• "I care about you, and I wanted to check in. How are you feeling?"

• "You have not seemed like yourself lately. I am here to listen. What has been going on?"

Ask and listen with care

Let the person talk at a pace that works for them. Ask open questions, listen closely, and avoid jumping straight into problem-solving.

You do not need to fix everything in the first conversation. It can help to focus on understanding what the person is going through.

Examples include:

• "That sounds really hard. Can you share more about what it has been like?"

• "I am really glad you shared this with me. What has been the hardest part lately?"

• "I am here with you, and I want to support you through this. How long have you been feeling this way?"

Ask directly if you are worried about suicide or self-harm

If you are worried about suicide or self-harm, ask plainly. Keep your tone calm and focus on listening. Research shows that asking people if they are suicidal does not increase suicidal thoughts or behaviors.

Examples include:

• "Have you been thinking about hurting yourself?"

• "Are you having thoughts about ending your life?"

• "Do you feel like you do not want to be alive anymore?"

• "Have you had moments where you have felt like you do not want to be alive anymore?"

If the person says yes, stay present and try to understand what they are going through.

You can gently ask a few direct follow-up questions to understand urgency and safety.

Examples include:

• "I am really glad you told me. Thank you for trusting me with this."

• "I am here with you. I am really sorry you are carrying this. Can we keep talking about this together?"

• "Are you thinking about doing something to hurt yourself today or sometime soon?"

• "What could help you feel safe right now?"

Avoid responses that can make things harder

Avoid minimizing the person's feelings, expressing judgment, or using guilt.

Examples to avoid include:

• "It is not that bad. You have so much to live for."

• "I cannot believe you would even think about killing yourself."

• "Think about how much it would hurt me and everyone who cares about you."

Help the person connect with more support

You do not need to have all the answers. Helping the person connect with more support is a strong next step.

You can offer to help the person contact a helpline, a mental health professional, or another trusted person.

Examples include:

• "I am here for you. I also think it could help to talk to a mental health professional."

• "Would it help if we called or texted a crisis line together?"

• "Would it be okay if we looked up options together? I can stay with you while you call or text."

If the person does not want to talk

If you believe the person can currently stay safe, respect their choice while leaving the door open.

Examples include:

• "It is okay if you are not ready to talk about it all right now. I am here, and we can take it one step at a time."

• "Would it feel easier to talk to someone else? We can bring in another person you trust."

If you believe the person may be in immediate danger, contact local emergency services or a helpline right away.

If immediate safety is a concern

Safety comes first. If the person may be in immediate danger, has a plan to harm themselves, cannot stay safe, or you do not feel safe, contact local emergency services.

You can also use a helpline for support and guidance. A trained listener can help you think through next steps and connect with additional support if needed.

Additional support resources:

• Find A Helpline

• American Foundation for Suicide Prevention: #RealConvo Guide: If Someone Tells You They’re Thinking About Suicide

• American Foundation for Suicide Prevention: #RealConvo Guide: How to Start (and Continue!) a Conversation About Mental Health

• National Institute of Mental Health: 5 Action Steps to Help Someone Having Thoughts of Suicide

Close with a next step

End the conversation by reinforcing care and making a concrete plan to check in again.

Examples include:

• "I am really glad we talked. I care about you. When would it work for you to talk again?"

• "Can I check in later today or tomorrow?"

• "What would be the best way for me to check in next time?"

If you cannot reach the person

If you cannot reach the person after a notification, try another way to contact them.

You can also consider reaching out to another trusted person who may be able to check on them.

If you believe the person may be in immediate danger, contact local emergency services right away.

Information for trusted contacts who are not users of OpenAI’s services

If a ChatGPT user asks you to be their trusted contact, OpenAI may receive your name, email address, and/or phone number from that user. If you are added as a trusted contact, OpenAI may also process information about your invitation and trusted contact status, such as whether the invitation was accepted, declined, expired, withdrawn, or removed.

OpenAI uses this information only to send and manage the invitation, record your response, administer the trusted contact relationship, deliver future safety notifications if you accept, and maintain the security and integrity of the feature. If a notification is sent, it is limited to letting you know that a safety concern has been detected and encouraging you to reach out.

OpenAI keeps this information only for as long as needed to operate the feature and manage the invitation or trusted contact relationship. Invitations expire if they are not accepted within the relevant period. OpenAI may keep some information for longer where needed for legal, security, or abuse-prevention reasons. For more information about how long OpenAI keeps personal data, please see the OpenAI Privacy Policy.

Depending on the country you’re in, OpenAI OpCo, LLC or OpenAI Ireland Limited is the controller for this processing. OpenAI may share information related to trusted contacts with its affiliates, service providers that help it operate and support the feature, and authorities or other third parties where required by law or necessary to protect rights, safety, and security. Your personal data may be processed outside the country where you live, including in the United States. In relevant countries, OpenAI’s legal basis for this processing is its legitimate interests in providing safety features and protecting the life, health, and safety of our users and others, including by supporting real-world outreach in serious safety situations. In exceptional cases involving a serious or imminent threat to someone’s life or physical safety, OpenAI may also rely on vital interests. You can object to this processing at any time by declining the invitation during the enrollment period or, if you have already accepted, by removing yourself as a trusted contact.

For more information about your privacy rights, international transfers, and how to contact OpenAI or its Data Protection Officer, please see the OpenAI Privacy Policy.

FAQ

Does being a trusted contact mean I am responsible for keeping the person safe?

No. Being a trusted contact does not make you responsible for the person's safety. It means you are someone they trust to check in and offer support.

Can I decline or stop being a trusted contact after I sign up?

Yes. You can decline during the 1-week enrollment period. If you have accepted the invitation and no longer wish to be a trusted contact, see: Declining or removing yourself as a trusted contact.

I received a notification that I was removed as a trusted contact. What does that mean?

There are different reasons that a person may be removed as a trusted contact. For example, you may not have been eligible to participate based on age, or the person who added you as their trusted contact may have removed you.