ChatGPT can now alert your Trusted Contact in critical moments

ChatGPT can now alert a Trusted Contact when it detects a user is going through a critical moment — and that shifts this AI from productivity tool to something with genuine real-world stakes. OpenAI is making a deliberate bet on active mental health support, which raises serious questions about privacy, liability, and where AI's role should end.

How we got here

Since its explosive launch in 2022, ChatGPT has become an unlikely confidant for millions of users dealing with anxiety, loneliness, and mental health struggles. OpenAI has long known its model receives the kind of disclosures that usually go to therapists or close friends. Regulatory and social pressure on AI platforms to take responsibility during high-risk interactions has intensified significantly over the past two years.

What the feature actually does

The new functionality lets users designate a Trusted Contact — a real person like a family member or friend — who ChatGPT can notify if a conversation signals the user may be in danger. According to Android Authority, the system is designed to intervene during critical moments, such as conversations involving self-harm or severe emotional distress. Crucially, it doesn't operate automatically without context: it requires prior setup by the user, preserving a meaningful degree of personal control. OpenAI has not yet disclosed the exact technical thresholds that trigger an alert or the precise notification mechanism involved.

What this really means

This move positions ChatGPT not just as a tool, but as a kind of digital safety net. The clearest winner is the vulnerable user who may lack easy access to professional support — the AI acts as a bridge to real human intervention. But it also opens a pointed debate: how far should an AI go in managing mental health crises? The risk of false positives and over-reliance on a language model to accurately detect genuine emergencies is not trivial and shouldn't be dismissed.

The broader industry impact

If this feature gains traction, competing models — Gemini, Claude, Copilot — will face pressure to build something equivalent. Regulators in both the EU and the US have spent months debating what obligations AI platforms should carry toward vulnerable users. OpenAI's move could become the de facto industry standard before any formal legislation arrives. For the tech industry at large, this marks an inflection point: AI is beginning to absorb functions that were previously considered exclusively human territory.

The real question is whether we're comfortable letting an AI make that call on our behalf.

Source: Android Authority