OpenAI is rolling out a new safety feature for ChatGPT. Users can now nominate a trusted contact. The AI assistant will alert that person if it detects signs of self-harm risk during a conversation.

The opt-in feature requires users to manually set up a Trusted Contact in their account settings. When ChatGPT identifies content suggesting a user may be at risk, it can send a notification to the designated person. OpenAI says the system uses existing moderation tools to detect concerning language.

How the Alert System Works

The trusted contact receives a message that the user appears to be struggling emotionally. The alert does not share the full conversation history. It includes a prompt encouraging the contact to reach out to the user directly.

This feature builds on ChatGPT's existing crisis response system. When a user expresses suicidal thoughts or self-harm intentions, the chatbot already provides links to mental health resources. The new option adds a human layer of support.

Users can change or remove their trusted contact at any time. The feature is available now on the web version of ChatGPT. A mobile rollout is expected soon.

Why This Matters

Mental health crises often go unnoticed by family and friends. A person may express distress to an AI assistant before telling anyone else. This feature creates a bridge between the digital conversation and real-world intervention.

For vulnerable users, the alert could make a critical difference. The feature may also reduce the burden on emergency services by enabling early support from someone the user trusts.

Privacy advocates have raised questions about data handling. OpenAI says alerts are triggered only when the system has high confidence of risk. The company also states that conversation data is not shared with the trusted contact beyond the alert itself.

The feature places new responsibility on both users and their nominated contacts. Experts suggest users discuss this option with their chosen person before enabling it.