ChatGPT Introduces Trusted Contact Feature for Mental Health Crisis Support.

OpenAI is launching Trusted Contact, an optional safety feature in ChatGPT that allows adult users to designate a trusted person who may be notified if the system detects signs of serious self-harm risk during a conversation.

The feature, announced on LinkedIn, was developed in collaboration with mental health experts from the American Psychological Association, OpenAI's Global Physician's Network, and the company's Expert Council on Well-Being and AI. It represents a shift toward connecting AI-detected crisis signals with real-world human support networks.

Users can opt into the feature and select one person in advance who serves as their designated contact. If ChatGPT's safety systems identify conversation patterns that suggest a serious self-harm concern, the designated person may receive a notification. The feature is designed to complement existing crisis resources and professional mental health care rather than replace them.

OpenAI emphasizes three core design principles: encouraging real-world connection and support, preserving user choice and privacy, and working alongside professional crisis intervention services. The feature is currently available only to eligible adult users, though OpenAI has not specified the exact criteria for eligibility.

The optional nature of Trusted Contact sets it apart from automated reporting systems used by some platforms. Users maintain control over whether to enable the feature and who to designate as their contact. This approach balances proactive safety measures with user autonomy, a persistent tension in content moderation and mental health interventions on digital platforms.

The privacy implications remain somewhat unclear from the initial announcement. OpenAI has not detailed what information is shared with the trusted contact when a notification is triggered, how long conversation data is retained for safety analysis, or whether users receive simultaneous alerts when their designated contact is notified.

The rollout comes as AI companies face growing scrutiny over how their systems handle sensitive topics and vulnerable users. While ChatGPT already provides crisis resource information when users express distress, Trusted Contact extends that response beyond on-screen text to potential real-world intervention.

This approach differs from recent product development at OpenAI, which has focused heavily on capability improvements. Recent updates include GPT-5.5 Instant with enhanced memory and Codex workspace integration for project management.

The feature also raises questions about how AI systems should balance user privacy with intervention in crisis situations. Mental health professionals have long debated duty-to-warn protocols, and Trusted Contact extends that discussion into conversational AI.

OpenAI has not provided a timeline for broader availability or indicated whether similar features might extend to enterprise ChatGPT deployments, where workplace mental health support presents different considerations. The company describes Trusted Contact as part of ongoing work to improve how ChatGPT responds in sensitive situations, suggesting further safety features may be in development.