On May 7, 2026, OpenAI unveiled a feature called Trusted Contact for ChatGPT accounts, designed to surface a human connection when conversations indicate possible self-harm. The tool lets adult users designate a trusted person — a friend or family member — who will be encouraged to check in if the system detects signs of distress. OpenAI says the aim is to provide an immediate, human-facing nudge alongside its existing automated safety measures.
How Trusted Contact Works
Users can add an adult trusted contact to their ChatGPT account. When the model or safety filters detect language suggesting suicidal thoughts or other serious self-harm risk, OpenAI’s workflow prompts the user to reach out to that contact. Simultaneously, the system can send an automated alert to the designated person via email, text, or an in-app notification. OpenAI emphasizes that these notifications are intentionally brief and do not include the detailed content of the conversation in order to protect the user’s privacy.
Human Review and Escalation
Trusted Contact complements OpenAI’s current incident handling, which mixes automated detection with human review. Safety flags trigger a notification that is reviewed by a human safety team; the company states it strives to complete reviews in under an hour. If reviewers determine there is a serious safety risk, the trusted contact alert becomes one of the responses the system will use to encourage human intervention.
Background: Lawsuits and Earlier Safeguards
OpenAI’s announcement comes amid legal pressures: several families have sued, alleging that interactions with ChatGPT contributed to suicides. Those cases put a spotlight on how AI platforms should respond when users express intent to harm themselves. Last September, OpenAI rolled out parental controls that allow parents to receive safety notifications for teen accounts — another layer of oversight that, like Trusted Contact, is optional.
Privacy, Optionality, and Limitations
OpenAI frames Trusted Contact as voluntary. Users must opt in and select a contact, and even when enabled, the measure has constraints. Alerts are summarized to preserve privacy and won’t relay the full conversation. Moreover, the existence of multiple ChatGPT accounts per person and the optional nature of parental controls and trusted contacts mean the feature cannot catch every high-risk interaction. The company acknowledges these limits while positioning Trusted Contact as one more tool in a broader safety toolkit.
Practical Implications for Users and Families
For people who opt in, Trusted Contact could provide a prompt that brings supportive intervention sooner than might otherwise occur. For relatives and friends, an unexpected alert can be the cue to reach out and potentially connect the person with professional help. However, the effectiveness depends on timely human response, the quality of the relationship between user and contact, and whether the contact knows how to respond to such alerts compassionately and constructively.
Where OpenAI Says It’s Headed
OpenAI frames Trusted Contact as part of ongoing improvements to how its systems handle moments of distress. The company says it will continue collaborating with clinicians, researchers, and policymakers to refine interventions and responses. That collaboration will likely shape future tweaks to detection sensitivity, notification practices, and privacy safeguards.
Conclusion
Trusted Contact adds a human-facing option to ChatGPT’s safety arsenal — a feature designed to nudge people toward support when automated systems detect risk. While it cannot eliminate all gaps and its optional nature limits coverage, it represents a step toward integrating human networks into AI safety responses. As OpenAI works with experts and regulators, the real-world impact of Trusted Contact will depend on adoption, responsible use, and how well alerts translate into timely, effective help.
Licensed under CC BY-ND 4.0 International
