OpenAI Launches Trusted Contact Feature to Prevent Self-Harm
1OpenAI has introduced a new safety feature called Trusted Contact, aimed at helping users who may be at risk of self-harm during conversations with its AI models. The feature allows adult users to designate a trusted person, such as a friend or family member, who can be alerted if the AI detects signs of distress or harmful intent. This addition is part of OpenAI’s ongoing efforts to improve safety and responsibility in AI interactions.
How the Trusted Contact Feature Works
With Trusted Contact, users can select someone they trust to be notified in case of concerning conversations. When the AI system identifies potential signs of self-harm or distress, it prompts the user to reach out to their trusted contact. If the user agrees, the system automatically sends an alert to that person through email, text, or in-app notification. The alert is designed to be simple and encouraging, prompting the trusted contact to check in with the user without sharing detailed conversation content to protect privacy.
OpenAI emphasizes that this feature is optional and designed to respect user privacy. The alerts are brief and do not disclose the specifics of the conversation, focusing instead on prompting support. The company also notes that users can have multiple accounts and choose different safety features for each, giving flexibility in how safety is managed across their AI interactions.
Background and Broader Safety Measures
This new safeguard builds on previous safety measures introduced last September, which gave parents oversight capabilities over their teens’ accounts. Those controls allowed parents to receive notifications if the system flagged a serious safety concern. Additionally, ChatGPT has long included automated prompts to encourage users to seek professional help if conversations veer toward self-harm or suicidal thoughts.
OpenAI has faced legal challenges from families claiming its chatbot encouraged or assisted loved ones in harming themselves. The company states that it uses a combination of automated detection and human review to handle risky situations. When a concerning interaction is identified, a safety team reviews the incident within about an hour, and if deemed necessary, triggers the Trusted Contact alert. This approach aims to intervene early and connect vulnerable users with support.
OpenAI highlights that Trusted Contact is part of a broader strategy to develop AI systems that assist during tough moments. The company plans to keep working with health professionals, researchers, and policymakers to improve these safety features and ensure responsible AI use.
What do you think?
It is nice to know your opinion. Leave a comment.