OpenAI has released a rough estimate of how many ChatGPT users globally may show signs of having a severe mental health crisis in a typical week. The company said Monday that it worked with experts around the world to make updates to the chatbot so it can more reliably recognize indicators of mental distress and guide users toward real-world support.

…

In a given week, OpenAI estimated that around 0.07 percent of active ChatGPT users show “possible signs of mental health emergencies related to psychosis or mania” and 0.15 percent “have conversations that include explicit indicators of potential suicidal planning or intent.”

OpenAI also looked at the share of ChatGPT users who appear to be overly emotionally reliant on the chatbot “at the expense of real-world relationships, their well-being, or obligations.” It found that about 0.15 percent of active users exhibit behavior that indicates potential “heightened levels” of emotional attachment to ChatGPT weekly. The company cautions that these messages can be difficult to detect and measure given how relatively rare they are, and there could be some overlap between the three categories.

OpenAI CEO Sam Altman said earlier this month that ChatGPT now has 800 million weekly active users. The company’s estimates therefore suggest that every seven days, around 560,000 people may be exchanging messages with ChatGPT that indicate they are experiencing mania or psychosis. About 1.2 million more are possibly expressing suicidal ideations, and another 1.2 million may be prioritizing talking to ChatGPT over their loved ones, school, or work.

…

OpenAI says it worked with over 170 psychiatrists, psychologists, and primary care physicians who have practiced in dozens of countries to help improve how ChatGPT responds in conversations involving serious mental health risks. If someone appears to be having delusional thoughts, the latest version of GPT-5 is designed to express empathy while avoiding affirming beliefs that don’t have basis in reality.

…

OpenAI says the medical experts reviewed more than 1,800 model responses involving potential psychosis, suicide, and emotional attachment and compared the answers from the latest version of GPT-5 to those produced by GPT-4o. While the clinicians did not always agree, overall, OpenAI says they found the newer model reduced undesired answers between 39 percent and 52 percent across all of the categories.