Over 1 Million Weekly ChatGPT Chats Show Signs Of Suicidal Thoughts: OpenAI Data

OpenAI revealed, in its blog post, on Monday, that over a million of the weekly ChatGPT users are engaged in conversations about suicidal ideas‍‌‍‍. The firm revealed that more than a million ChatGPT users weekly are sending messages which include "explicit indicators of potential suicidal planning or intent". This figure represents a frightening and escalating coupling of AI engagement and mental health problems.

Among more than 800 million active users each week, ChatGPT is involved in "more than a million" conversations related to suicide ideation every week. According to The Guardian, this statement is one of the clearest and most direct indications from the large AI corporation about how much artificial intelligence might exacerbate mental health problems.

Moreover, the AI chatbot is said to have a weekly interaction with some 560,000 individuals who exhibit symptoms of mania or psychosis, and a similar number of users show "increased emotional attachment to ChatGPT".

OpenAI suggests that such ChatGPT conversations are "extremely rare" and hence difficult to quantify. Nevertheless, the company estimated that hundreds of thousands of people are affected by these issues every week.

The firm said on Monday that it had collaborated with professionals from all over the world to upgrade the chatbot so that it would be more capable of recognising mental distress cues and providing in-person resource information to the users.

In its blog post, OpenAI claimed that its most recent GPT-5 is safer for users and less prone to unwanted behaviours, as evidenced by the model evaluation comprising more than 1,000 chats about suicide and self-harm, in which the company did not immediately respond to a request for comment.

“Our new automated evaluations score the new GPT‑5 model at 91% compliant with our desired behaviours, compared to 77% for the previous GPT‑5 model,” the firm's site reads.

OpenAI claims that it consulted more than 170 mental health professionals from its Global Physician Network of healthcare for its latest work on ChatGPT. These healthcare professionals observed that ChatGPT "is more appropriate and consistent in its responses than its earlier versions".

“As part of this work, psychiatrists and psychologists examined over 1800 responses of models on topics of serious mental health issues and compared the new GPT‑5 chat model's responses to those of the previous models,” OpenAI said.

While noting such improvements, mental health experts still maintain that an AI chatbot cannot fully replace therapy provided by human experts. They point to occurrences of "sycophancy," where an AI system might support negative views, and warn that susceptible users may decide to use chatbots rather than seek help from people.

This disclosure arrives amid heightened scrutiny by regulators such as the US Federal Trade Commission and other agencies, who are looking into how internet companies measure their impacts on the well-being of children and teenagers. OpenAI stresses that the figures represent the gravity of the mental health issues that people raise in the conversations rather than implying that ChatGPT is the source of the ‍‌‍‍‌‍‌‍‍‌distress.