OpenAI Adds 'Trusted Contacts' Feature to ChatGPT to Address Self-Harm Risks in Conversations

OpenAI recently launched the 'Trusted Contacts' feature for ChatGPT. When self-harm-related content appears in user conversations, the system encourages adult users to reach out to others and sends reminders to pre-set trusted contacts to promote communication.

Previously, OpenAI faced lawsuits over ChatGPT being accused of encouraging or assisting users in making suicide plans. Currently, the platform uses automatic review plus manual verification to handle potential self-harm incidents; after a reminder is triggered, information is forwarded to the manual safety team, aiming to complete reviews within one hour.

If a user is deemed at serious risk, ChatGPT sends notifications to trusted contacts via email, SMS, or push alerts. The content is brief and excludes conversation details to protect privacy while reminding contacts to check on the user.