Hundreds of Thousands of ChatGPT Users Show Signs of Mental Crisis Weekly
▼ Summary
– OpenAI has released its first estimate of ChatGPT users showing signs of severe mental health crises, collaborating with experts to improve the chatbot’s response to mental distress.
– Recent incidents include hospitalizations, divorces, and deaths linked to intense ChatGPT conversations, with concerns raised about AI-fueled delusions and paranoia.
– Approximately 0.07% of weekly active users show possible signs of psychosis or mania, while 0.15% exhibit indicators of suicidal planning or intent.
– An additional 0.15% of users display emotional reliance on ChatGPT that harms real-world relationships or obligations, with potential overlap among these categories.
– Based on 800 million weekly users, these estimates translate to about 560,000 people possibly experiencing psychosis or mania and 2.4 million showing suicidal ideation or excessive attachment weekly.
Navigating the intersection of artificial intelligence and mental health, new data from OpenAI reveals a startling trend among ChatGPT users. For the first time, the company has provided a rough estimate of how many individuals using the chatbot globally may exhibit signs of severe psychological distress each week. Working alongside international experts, OpenAI has rolled out updates aimed at helping the system more reliably identify indicators of mental health crises and direct users toward appropriate real-world support.
Recent months have seen a troubling pattern emerge, with some individuals reportedly facing hospitalization, divorce, or even death following prolonged and intense interactions with ChatGPT. Family members of affected users have claimed the AI system amplified their loved ones’ delusions and paranoia. Mental health professionals have voiced serious concerns about this phenomenon, often termed AI psychosis, but until now, comprehensive data on its prevalence has been lacking.
OpenAI’s analysis indicates that approximately 0.07% of active ChatGPT users weekly display possible signs of mental health emergencies linked to psychosis or mania. Additionally, about 0.15% of users engage in conversations containing explicit indicators of potential suicidal planning or intent. The company also examined cases where individuals appear to develop an excessive emotional reliance on the chatbot, potentially at the cost of real-world relationships, personal well-being, or daily responsibilities. Their findings show roughly 0.15% of active users weekly exhibit behaviors suggesting a heightened emotional attachment to ChatGPT.
Because CEO Sam Altman recently stated that ChatGPT now serves around 800 million weekly active users, these percentages translate to significant numbers. Each week, an estimated 560,000 people may be communicating with ChatGPT in ways that suggest they are experiencing mania or psychosis. Furthermore, up to 2.4 million additional users could be expressing suicidal thoughts or prioritizing conversations with the AI over interactions with family, school, or work.
To address these risks, OpenAI collaborated with more than 170 psychiatrists, psychologists, and primary care physicians from dozens of countries. Their input has helped refine how ChatGPT handles dialogues involving serious mental health dangers. The latest iteration, GPT-5, is engineered to respond with empathy when users express delusional beliefs, while carefully avoiding validation of ideas not grounded in reality. For instance, if a user claims that planes flying overhead are targeting them, ChatGPT will acknowledge their feelings but clarify that no external aircraft or force has the ability to steal or implant thoughts.
(Source: Wired)
