Some ChatGPT messages may be related to mental health emergencies

New estimates of the number of ChatGPT users showing signs of possible mental health emergencies, including psychosis or suicidal thoughts, have been released by OpenAI, the BBC reports.

The company noted that it estimates that approximately 0.07 percent of ChatGPT users active in a given week may have messages related to mental health emergencies, indicating that these cases are "extremely rare." However, since ChatGPT's weekly active users recently reached 800 million, this tiny percentage could represent hundreds of thousands, according to OpenAI CEO Sam Altman.

The company also estimated that 0.15 percent of ChatGPT users engaged in conversations that contained "clear indicators of potential suicide planning or intent."

In its statement regarding the predictions, the company noted that with recent updates to the chatbot, ChatGPT is designed to "safely and empathically respond to possible signs of delusions" and detect "indirect signs of potential self-harm or suicide risk."

"On a platform with hundreds of millions of users, 0.07 percent may seem like a small percentage, but it could actually be quite large," said Jason Nagata, Associate Professor at the University of California.

"AI can expand access to mental health support and support mental health in some ways, but we must be aware of the limitations," Nagata said.