OpenAI Reveals One Million Weekly ChatGPT Users Exhibit Suicidal Indicators

OpenAI has disclosed that approximately 0.15% of its 800 million weekly active ChatGPT users, equating to over one million people, engage in conversations indicating potential suicidal thoughts or plans. The company has collaborated with more than 170 mental health professionals from its Global Physician Network to improve the chatbot's responses to mental health crises, psychosis, mania, and emotional over-reliance. The newly released GPT-5 model demonstrates significant improvements by reducing unsafe or non-compliant responses by up to 80%, achieving 91-92% compliance with safety protocols, and better handling complex or lengthy dialogues. OpenAI emphasizes that ChatGPT is not a therapist but aims to recognize signs of distress and direct users to professional help, including crisis helplines, and occasionally encourages breaks during emotionally charged sessions. The company faces scrutiny following lawsuits alleging that it downgraded suicide prevention safeguards to increase user engagement, highlighting the industry's challenges in balancing user safety with market goals. Despite these concerns, OpenAI continues to enhance safety mechanisms to mitigate harm while acknowledging the significant emotional and psychological reliance users place on AI.