On May 7, 2026, OpenAI announced a new ChatGPT safety feature called "Trusted Contact", which lets a user nominate one trusted adult who can receive a short notification if the conversation shows signs of serious self-harm risk[1]. The feature combines automated detection with a small team of trained human reviewers, and the alert itself contains no transcript or detailed conversation content[1][2].

Overview and eligibility

Trusted Contact is an opt-in setting that users turn on from inside ChatGPT[1][3]. The nominee must be an adult — 18 or older worldwide, and 19 or older in South Korea — and only one Trusted Contact can be active at a time[1][4]. The nominee receives an invitation by email, text, or in-app message that explains the role, and they have one week to accept before the protection becomes active[1][4]. If they decline or do not respond in time, the user can pick a different adult, and they can also change or remove the contact later from settings[1][4].

How notifications are decided

Trusted Contact does not rely on simple keyword matching. An automated classifier first flags conversations that may indicate a serious self-harm concern, after which ChatGPT encourages the user to reach out to their Trusted Contact and offers suggested ways to start that conversation[1]. In parallel, a small team of specially trained reviewers looks at the situation and decides whether a notification should actually be sent[1][2]. OpenAI says it aims to complete each safety review in under one hour[1][2]. When a notification is sent, it is brief and does not include the contents of the conversation, in order to protect the user's privacy[1][2].

Background and how it fits with existing safeguards

Since November 2025, OpenAI has faced multiple lawsuits from families of people who died by suicide after long conversations with ChatGPT[2]. Trusted Contact is part of a broader response to those concerns and follows the parental controls that OpenAI introduced in September 2025, which allow parents to keep a degree of oversight over their teens' accounts[2]. Where parental controls are limited to minors, Trusted Contact extends a similar idea to adults and lets each person choose who is alerted[1][2]. ChatGPT has long pointed users toward professional help when conversations turn toward self-harm; Trusted Contact adds an extra, person-to-person layer[2]. The company also acknowledges practical limits: the feature is optional, and a user can simply maintain a separate ChatGPT account where it is not enabled[2].

Summary

Trusted Contact gives a ChatGPT user the option to designate one trusted adult who can be notified, with no transcript shared, when the system flags a serious self-harm concern that human reviewers confirm. It is not a complete solution, but the combination of opt-in design, human review, and minimal information sharing reflects an attempt to balance safety with privacy, and it makes a quiet but meaningful bridge to a real person available as a built-in part of ChatGPT.

Source: https://openai.com/index/introducing-trusted-contact-in-chatgpt/

Source: https://techcrunch.com/2026/05/07/openai-introduces-new-trusted-contact-safeguard-for-cases-of-possible-self-harm/

Source: https://www.androidauthority.com/chatgpt-trusted-contact-3665049/

Source: https://www.medianama.com/2026/05/223-openai-trusted-contact-chatgpt-self-harm-risks/