Table of Contents
ChatGPT Trusted Contact: What It Does and How to Set It Up
ChatGPT trusted contact alerts one adult you pick if a chat signals serious suicide risk. Who can use it, how to set it up, what it shares, why a plan matters.
ChatGPT trusted contact is an optional safety feature that lets a user pick one adult who may get a short alert if ChatGPT’s systems and trained human reviewers decide a conversation shows a serious suicide risk. OpenAI launched it in May 2026. It has one large catch for families: it is only for users 18 and older, so it is a tool for your college freshman, not your 15-year-old.
It also has a quieter catch. When NPR tested it in September 2026, a brand-new account was never told the feature existed, and ChatGPT at first suggested the reporter might be “thinking of another app.” A safety switch nobody knows about does nothing. This guide covers what the feature does, the exact setup steps, what it shares and what it does not, and why the plan around it matters more than the toggle.
Key Takeaways
- Trusted Contact is for personal ChatGPT accounts held by adults 18 and older (19 in Korea). Teens under 18 are covered instead by parental controls and their safety notifications.
- Setup is in Settings > Trusted contact. The contact must be 18 or older and accept an invitation within one week. Each account gets one contact.
- A notification goes out only after automated systems flag a conversation and trained human reviewers judge it a serious safety concern. It says suicide came up in a concerning way. No chat details or transcripts are shared.
- It does not work in ChatGPT Voice, or in Business, Enterprise or Edu workspaces.
- OpenAI says plainly that it is “not an emergency service.” Treat it as one strand of a family safety plan, never the plan itself.
If you are worried about someone right now: In the US, call or text 988 (Suicide & Crisis Lifeline) or text HOME to 741741 (Crisis Text Line). Outside the US, find a local line at findahelpline.com. In an emergency, call 911.
What ChatGPT trusted contact is, and who can use it
Trusted Contact is a pre-arranged alert. You name one person; they agree in advance; if a later conversation crosses a high threshold of concern about suicide, ChatGPT may tell them to check in on you. According to OpenAI’s help center, it “is not an emergency service, a crisis response system, or a substitute for mental health care.”
Who can use it, per OpenAI:
- Account holders: adults 18 or older, with personal ChatGPT accounts, in most countries. In Korea the age is 19.
- Contacts: must also be 18 or older and must accept the invitation. They do not need a ChatGPT account.
- Not available: in ChatGPT Voice, or in shared workspaces such as ChatGPT Business, Enterprise or Edu.
That age rule changes how parents should think about it. For a teen under 18, the equivalent protection is a linked teen account through parental controls, which can send a parent limited safety notifications. For an 18-year-old, parental controls end. OpenAI’s parental controls documentation says that when ChatGPT identifies a linked account as belonging to someone 18 or older, parental controls are removed and “teen safety notifications stop for the account.” Trusted Contact is the adult option that can fill that gap, if your young adult chooses it.
This matters because young adults are the heaviest users of chatbots for mental health. A nationally representative survey in JAMA Pediatrics found 19.2% of US 12- to 21-year-olds had used chatbots for mental health advice in 2025, and 18- to 21-year-olds were far more likely to do so than 12- to 14-year-olds.
How Trusted Contact works, step by step
OpenAI describes four stages.
1. You invite one person. You provide their name and email (required) and, ideally, a phone number. Invitations may arrive by email, text, WhatsApp or in-app message.
2. They accept within one week. The invitation explains the role. If they decline, don’t respond or are ineligible, you are told so you can choose someone else.
3. Automated monitoring, then human review. If automated systems suggest you may be talking about suicide in a way that signals a serious concern, “trained human reviewers may review the conversation.” If a conversation is flagged for review, ChatGPT tells you it may notify your contact. If you don’t get that message, the conversation was not sent for review and your contact will not be notified.
4. A brief notification. If reviewers decide the situation may be serious, your contact gets a short message by email, text, WhatsApp or app saying that you discussed suicide in a way that may indicate a serious safety concern, and encouraging them to reach out. “Your chat details or transcripts are not shared.”
Here is the engineering point worth understanding. This is a two-stage detector: a fast automated classifier that scans for risk, followed by slower human judgment. The design trades speed for fewer false alarms. A human review step means notifications are less likely to fire on a dark joke, but it also means the alert is not instant, and the contact may not be reachable right away. OpenAI says so: “No system is perfect.”
How to set it up
From the OpenAI setup guide:
- Go to Settings > Trusted contact.
- Select Add contact.
- Enter the person’s Name, Email (required) and Phone (optional but recommended).
- Select the consent checkbox.
- Select Save changes.
The person then gets an invitation and has one week to accept. To change contacts, you remove the current one and add the new person, who must accept a fresh invitation. If you remove someone, they get a brief notice. You can ask ChatGPT whether a contact is set up or an invite is pending, but you cannot add or change one inside a conversation.
OpenAI recommends talking to the person first, and offers sample wording such as: “If you ever got a message saying I might be having a hard time, it would mean I could use a check-in. Would you feel comfortable reaching out to me if that happened?”
Trusted Contact vs parental controls vs a family plan
| Trusted Contact | Parental controls (teen accounts) | A family safety plan | |
|---|---|---|---|
| Who it covers | Users 18+ | Linked teens under 18 | Everyone in the family |
| Who decides to turn it on | The user | Parent or teen sends the invite; both must link | The family, together |
| What triggers an alert | Human reviewers judge a serious suicide concern | Trained reviewers identify serious self-harm concern or certain disordered-eating signals; some account deactivations | Anyone noticing warning signs |
| Shares chat content? | No | No | Whatever the person chooses to say |
| Works in voice mode? | No | Not stated for notifications | Yes, it’s people |
| Real-time? | No | No, “not real-time monitoring” | As fast as someone calls |
For teens, our explainer on ChatGPT parental controls walks through linking accounts. If you are unsure whether ChatGPT treats your child as an adult, see how ChatGPT’s age prediction works.
Why the plan matters more than the toggle
NPR’s September 2026 investigation is the clearest case for this. A producer set up a new account; onboarding never mentioned Trusted Contact. When asked, ChatGPT initially didn’t recognize the feature. NPR, citing OpenAI’s fall 2025 data, estimated that more than a million weekly users discuss suicide and self-harm with ChatGPT. OpenAI reported “good adoption” but gave no numbers.
The experts NPR interviewed agreed on the core issue. Johns Hopkins’ Holly Wilcox said it would be ideal to offer the feature “to all users when they are not yet in an emergency.” Ursula Whiteside of Now Matters Now found that people wanted their contact chosen in advance, while thinking clearly. And the University of Alabama’s Jennifer Lockman said she wants research on the feature’s impact before anyone concludes it saves lives.
Three limits make the plan essential:
- It only sees ChatGPT. Distress expressed elsewhere, or out loud, is invisible to it.
- It’s off in voice mode, which many people use when upset.
- It can’t act. It sends a message. A person has to respond.
The APA’s November 2025 health advisory puts it bluntly: “Relying solely on an app during a mental health emergency can be dangerous.”
How to build the family plan around it
Have the conversation before the setup
If your young adult is willing to add you, or someone else, sit down first. Ask what kind of check-in would feel supportive: a call, a text, a visit. OpenAI’s own guidance suggests agreeing on this in advance. If they would rather choose a sibling, aunt or close friend, respect it. A contact they trust beats a contact they resent.
Decide what the contact will do if an alert comes
OpenAI’s guidance for trusted contacts is sound: find a private moment, lead with care, and if you are worried, ask directly about suicide. It notes that research shows asking does not increase suicidal thoughts or behaviors. Help the person contact a crisis line or professional, and agree on when you will check in again. If you cannot reach them and believe they are in danger, call emergency services.
Learn the warning signs ChatGPT can’t see
The National Institute of Mental Health lists talking about wanting to die or being a burden, feeling hopeless or trapped, withdrawing from friends, giving away belongings, mood swings and changes in sleep or eating. Suicide is rarely about one thing, so watch the whole picture.
Save the human numbers
Put 988 in your young adult’s phone and yours. If they study or live abroad, look up their local line on findahelpline.com together.
What not to do: make the toggle a condition of support
Demanding that an 18-year-old add you as a trusted contact, or else, can push them to keep struggles private. Offer it, explain why, and accept their choice of person.
What to Watch For Over the Next 3 Months
- Week 4: If your young adult set up a contact, confirm the invitation was accepted within the one-week window. Ask the contact if they know what they’d do on receiving an alert.
- Month 2 red flags: The NIMH warning signs, plus heavy late-night chatbot use, hiding app use or withdrawing from friends. Don’t wait for an alert; reach out and contact a professional. Call 988 or 911 if there is immediate danger.
- Month 3 self-check: Is the contact’s phone number and email still current? Does your young adult know at least one crisis line by heart, and one in-person resource such as campus counseling?
Frequently Asked Questions
Can I set up ChatGPT Trusted Contact for my 15-year-old?
No. Trusted Contact is only for users 18 and older. For under-18s, link accounts through ChatGPT’s parental controls, which can send you limited safety notifications after human review. Those notifications stop when the account is identified as belonging to an adult.
Will a trusted contact see my child’s chats?
No. OpenAI says the notification gives only the general reason, that suicide came up in a potentially concerning way, and encourages the contact to check in. Chat details and transcripts are not shared.
Does Trusted Contact work if my child uses ChatGPT by voice?
No. OpenAI says Trusted Contact is not currently available in ChatGPT Voice. That is a real gap, since people often talk rather than type when upset.
Is Trusted Contact enough to keep my young adult safe?
No. OpenAI calls it “not an emergency service,” and the APA warns that relying solely on an app in an emergency can be dangerous. It is one backup. In a crisis, call or text 988, text HOME to 741741, use findahelpline.com, or call 911.
What if I get a trusted contact notification?
Reach out soon, privately and calmly. Lead with care, ask how they are, and ask directly if you are worried about suicide. Help them contact a crisis line or professional, and plan a follow-up. If you can’t reach them and fear for their safety, call emergency services.
About the author
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
Sources
- OpenAI Help Center. (2026). “Trusted contacts in ChatGPT.” OpenAI. https://help.openai.com/en/articles/20001105-trusted-contacts-in-chatgpt
- OpenAI Help Center. (2026). “Adding and managing a trusted contact.” OpenAI. https://help.openai.com/en/articles/20001194-adding-and-managing-a-trusted-contact
- OpenAI Help Center. (2026). “Being someone’s trusted contact.” OpenAI. https://help.openai.com/en/articles/20001195-being-someones-trusted-contact
- OpenAI Help Center. (2026). “Managing parental controls in ChatGPT.” OpenAI. https://help.openai.com/en/articles/12315553
- Chatterjee, R. (2026, September 14). “NPR looked for ChatGPT’s new crisis feature. It’s hard to find.” NPR. https://www.npr.org/2026/09/14/nx-s1-5943118/chatgpt-has-a-feature-to-help-those-in-crisis-but-where-can-you-find-it
- McBain, R. K., Cantor, J. H., et al. (2026). “AI Chatbot Use and Disclosure for Mental Health Among US Adolescents and Young Adults.” JAMA Pediatrics. https://doi.org/10.1001/jamapediatrics.2026.2015
- American Psychological Association. (2025, November). “APA Health Advisory on the Use of Generative AI Chatbots and Wellness Applications for Mental Health.” APA. https://www.apa.org/topics/artificial-intelligence-machine-learning/health-advisory-ai-chatbots-wellness-apps-mental-health.pdf
- National Institute of Mental Health. “Warning Signs of Suicide.” NIH. https://www.nimh.nih.gov/health/publications/warning-signs-of-suicide
