Over a million people discuss suicide with ChatGPT weekly, says OpenAI

ChatGPT maker OpenAI released new data this week showing that about 0.15% of its 800 million weekly active users—over 1 million people a week—have conversations with the chatbot that contain signs of suicide planning or self-harm intent.

A similar percentage form strong emotional attachments to the AI, while hundreds of thousands show signs of psychosis or mania in their interactions.

OpenAI shared these findings as part of a broader update on improving ChatGPT’s responses to mental health crises. The company currently faces a lawsuit involving a teenager’s suicide and warnings from state attorneys general.

CEO Sam Altman claims the GPT-5 improvements have mitigated serious mental health risks, but experts still question the long-term impact.