OpenAI says over a million people talk to ChatGPT about suicide weekly

OpenAI is released New information Monday explains how many of ChatGPT’s users suffer from mental health issues, and talk to AI chatbots about it. The company said that 0.15% of ChatGPT’s active users in a given week had “conversations that included clear indicators of possible suicidal planning or intent.” Given that ChatGPT has over 800 million weekly active users, that translates to over a million people per week.

The company said the same percentage of users showed “high levels of emotional attachment to ChatGPT” and that tens of thousands showed signs of psychosis or mania in their weekly conversations with the AI ​​chatbot.

OpenAI says such conversations on ChatGPT are “extremely rare” and therefore difficult to measure. That said, OpenAI estimates that these issues affect hundreds of thousands of people every week.

OpenAI shared the information as part of a broader announcement about recent efforts to improve how models respond to users with mental health issues. The company claims its latest work on ChatGPT involved consulting more than 170 mental health experts. OpenAI said these clinicians observed that the latest version of ChatGPT “responds more appropriately and consistently than previous versions.”

In recent months, several stories have come to light about how AI chatbots can Adversely affect users Struggling with mental health challenges. Researchers have found this before AI chatbots can lead some users down a confusing rabbit holeMainly reinforces dangerous beliefs through sycophantic behavior.

Addressing mental health concerns in ChatGPT is fast becoming an existential problem for OpenAI. Currently the company is A case has been filed against the parents of a 16-year-old boy who had concealed her suicidal thoughts with ChatGPT in the weeks leading up to her suicide. The state attorneys general of California and Delaware — which could block the company’s planned restructuring — have also warned OpenAI that it The youth must be protected who use their products.

Earlier this month, OpenAI CEO Sam Altman made a claim in one Posted on X The company has been able to “mitigate serious mental health issues” in ChatGPT, though he didn’t give specifics. The data shared Monday appears to back up that claim, though it raises broader questions about how widespread the problem is. Still, Altman says OpenAI will loosen some restrictions, even allowing adult users to get started Having a sexy conversation With AI chatbots.

In Monday’s announcement, OpenAI claimed that the recently updated version of the GPT-5 responded with “desired responses” to about 65% more mental health problems than the previous version. In tests evaluating AI responses around suicidal conversations, OpenAI says its new GPT-5 model is 91% consistent with the company’s desired behavior, compared to 77% for the previous GPT-5 model.

The company also says that its latest version of GPT-5 also better retains OpenAI’s protections over long conversations. OpenAI previously flagged that its protections were less effective on long conversations.

On top of these efforts, OpenAI says it is adding New assessment ChatGPT to measure some of the most serious mental health challenges faced by users. The agency said its baseline safety tests for AI models will now include benchmarks for psychological dependence and non-suicidal mental health emergencies.

OpenAI has recently rolled out more Controls for parents Children using ChatGPT. The company said it is developing an age prediction system to automatically identify children using ChatGPT and impose a stricter security system.

Still, it’s unclear how enduring the mental health challenges around ChatGPT will be. While GPT-5 appears to be an improvement over previous AI models in terms of security, there still appears to be a portion of ChatGPT’s responses that OpenAI considers “undesirable”. OpenAI still makes its older and less-secure AI models, including GPT-4o, available to its millions of paying customers.