00:00A new and alarming report from OpenAI has revealed that around 1.2 million users have engaged in conversations with ChatGPT containing signs or intentions of self-harm.
00:16The review, which analysed 1,000 conversations, found that the chatbot responded to 91% of these messages without warning users or discouraging them.
00:27OpenAI said the findings reflect repeated exposure to AI-generated content that may worsen mental distress.
00:35The company has since strengthened parental controls and added safety measures, including directing sensitive chats to safer models and linking users to mental health helplines.
Be the first to comment