For millions of users, AI chatbots are no longer just tools for writing emails or answering questions. They are companions during lonely nights, sounding boards during emotional spirals, and increasingly, spaces where deeply personal conversations unfold.
That growing emotional connection between users and AI is now forcing tech companies to rethink what responsibility looks like when conversations take a darker turn.
Taking note of this trend, OpenAI announced a new ChatGPT safety feature called “Trusted Contact”, designed to help users during moments of emotional distress. The optional feature allows adult users to add a trusted person, such as a close friend, partner, or family member, who can be contacted if OpenAI’s systems detect signs of possible self-harm during a conversation.
The move comes as OpenAI continues to face mounting pressure over chatbot safety, particularly after several lawsuits accused ChatGPT of encouraging emotionally vulnerable users towards self-harm or suicide.
OpenAI’s Trusted Contact feature: How does it work
The Trusted Contact tool is built into OpenAI’s existing safety systems, which already monitor certain high-risk conversations involving suicidal thoughts or self-harm. According to the company, ChatGPT uses a mix of automated systems and human reviewers to identify potentially dangerous interactions.
When conversations trigger safety concerns, the case is forwarded to OpenAI’s internal review team. The company says a human reviewer reviews every serious alert and aims to review such incidents quickly.
“We strive to review these safety notifications in under one hour,” OpenAI said.
If the company determines there is a serious safety risk, ChatGPT may encourage the user to contact their trusted person directly. In some situations, OpenAI’s systems can also send automated notifications to the chosen contact via text message, email, or an in-app alert.
OpenAI states, “We developed Trusted Contact with guidance from clinicians, researchers, and organizations that specialize in mental health and suicide prevention. This work is informed by our Global Physicians Network, a network of more than 260 licensed physicians across 60 countries, and our Expert Council on Well-Being and AI. We also worked closely with external organizations, including the American Psychological Association.”
Importantly, the notification does not reveal the details of the conversation itself. Instead, it simply informs the trusted person that the user may need support and encourages them to check in.
OpenAI says the approach is designed to offer intervention without significantly compromising user privacy.
The company also noted that Trusted Contact is entirely optional and available only to users who choose to activate it within their account settings.
Why is it important
The launch reflects a broader challenge facing AI companies as chatbots become increasingly conversational, emotionally aware,e and embedded in daily life.
Over the past year, OpenAI has faced a wave of criticism and legal complaints from families who allege ChatGPT failed to respond appropriately when users discussed self-harm. Some lawsuits claim the chatbot reinforced dangerous thoughts instead of discouraging them.
The company had already introduced safeguards before this latest update. Last September, OpenAI rolled out parental supervision tools for teenage users, allowing parents to receive alerts if the platform believed their child might be facing a “serious safety risk”. ChatGPT also routinely suggests professional mental health resources when conversations veer towards self-harm.
Still, the system has limitations. Users can maintain multiple ChatGPT accounts, and both parental controls and Trusted Contact settings must be manually enabled.
“Trusted Contact is part of OpenAI’s broader effort to build AI systems that help people during difficult moments,” the company wrote in its announcement. “We will continue to work with clinicians, researchers, and policymakers to improve how AI systems respond when people may be experiencing distress.”
As AI grows more personal, companies like OpenAI are increasingly being judged not only by what their systems can do, but also by how safely they respond when users are at their most vulnerable.
