OpenAI has rolled out a new “Trusted Contact” safeguard for ChatGPT users, allowing the AI platform to suggest reaching out to a pre-selected trusted person when conversations indicate potential self-harm or emotional crisis risks. The move comes as AI companies face increasing scrutiny over mental health safeguards and responsible AI deployment.
TL;DR
- OpenAI introduced a “Trusted Contact” feature for ChatGPT users facing emotional distress or possible self-harm risks.
- Users can voluntarily add a trusted person to their account settings.
- ChatGPT may encourage users to contact that person during high-risk conversations.
- The feature builds on OpenAI’s broader AI safety and mental health protections.
- OpenAI says the system is designed to support users without replacing professional mental health services.
OpenAI is strengthening its AI safety framework with a new “Trusted Contact” feature designed to intervene during sensitive conversations involving possible self-harm.
The feature allows users to designate a trusted individual who can be recommended as a support contact if ChatGPT detects signs of emotional crisis, distress, or suicidal ideation during conversations.
The safeguard is optional and requires users to proactively select a trusted contact through account settings. Once enabled, ChatGPT may encourage users to reach out to that person during conversations that trigger OpenAI’s mental health safety systems.
OpenAI clarified that the feature does not automatically alert or notify the trusted individual. Instead, it acts as a guided intervention mechanism aimed at encouraging human connection during vulnerable moments.
The company said the initiative was developed with input from mental health experts and aligns with its broader efforts to make conversational AI systems safer and more responsible.
In a statement shared with media outlets, OpenAI said it wants ChatGPT to respond appropriately when users are in distress while recognizing that ChatGPT is not a substitute for professional care.
The company added that the safeguard complements existing measures that already encourage users to contact emergency services, crisis hotlines, or professional support organizations when high-risk content is detected.
The announcement arrives amid growing concerns surrounding the emotional dependency users can develop with AI chatbots. Researchers and policymakers have increasingly warned about the psychological risks tied to highly conversational AI systems, especially as users spend more time interacting with them for companionship, therapy-like discussions, and emotional support.
Several AI firms, including OpenAI, Anthropic, Google, and Meta, have recently expanded their AI safety policies around mental health, manipulation risks, and harmful outputs.
OpenAI has been under particular scrutiny due to ChatGPT’s massive global reach. The company reportedly serves hundreds of millions of weekly active users across consumer and enterprise products.
The Trusted Contact feature also reflects a broader shift in the AI industry toward introducing human-in-the-loop safety systems. Instead of relying solely on automated moderation, companies are increasingly exploring mechanisms that reconnect vulnerable users with real-world human support.
According to OpenAI, the system has safeguards to minimize false positives and avoid overreacting to casual conversations. The company emphasized that users remain fully in control of whether they add a trusted contact and whether they follow ChatGPT’s recommendations.
Topics for more insights:
The rollout also comes as regulators worldwide push for stricter oversight of AI systems interacting with minors, vulnerable individuals, and users experiencing mental health challenges.
While OpenAI maintains that ChatGPT is not designed to function as a therapist or crisis counselor, the company acknowledged that many users naturally turn to conversational AI during emotionally difficult situations.
The Trusted Contact feature appears to be OpenAI’s latest attempt to balance accessibility, user autonomy, and responsible AI deployment as AI assistants become increasingly integrated into daily life.
Join The Discussion