In a shocking revelation, it has appeared that over a million people around the world use ChatGPT each week to talk about suicide. According to the company’s latest analysis, around 0.15% of its roughly 800 million weekly active users show signs of potential suicidal planning or intent. Another 0.07% of users display symptoms linked to psychosis or mania, while around 560,000 users appear to have formed an unusually strong emotional attachment to the chatbot, sometimes valuing its company over real-world interactions.
As a result, acknowledging the gravity of these findings, OpenAI has made major updates to how ChatGPT handles mental health-related conversations. The company worked with more than 170 mental health professionals to refine the GPT-5 model, improving how it identifies and responds to users in distress. OpenAI reports a 65% increase in positive responses to mental health crises and a 91% compliance rate with its internal safety standards during evaluations involving suicidal conversations.
The updated ChatGPT is now designed to better recognise sensitive topics and gently redirect users toward crisis hotlines or professional help. It also encourages users to take breaks during prolonged sessions and includes parental control options to limit exposure for younger users.
Despite progress, OpenAI admits the challenge isn’t fully solved. Vulnerable users may ignore interventions or avoid seeking help outside the chatbot. The company says it will continue collaborating with mental health experts to enhance ChatGPT’s ability to detect and support people in crisis while ensuring conversations remain safe and empathetic.
