OpenAI says ChatGPT is handling with thousands of suicidal users each week

OpenAI shared new data on Monday showing that a small but significant number of ChatGPT users are turning to the chatbot for mental health struggles, including suicidal thoughts. The company said that about 0.15% of weekly active users show signs of suicidal planning or intent.

Since ChatGPT has more than 800 million weekly users, this means over one million people may be discussing serious mental health issues through the platform.

The report also found that a similar share of users show emotional dependence on the chatbot, while hundreds of thousands of weekly conversations may include signs of psychosis or mania. Although these cases are rare, OpenAI said they are still a major concern because of the platform’s large user base.

OpenAI Works to Improve Responses

To make ChatGPT safer for people in distress, OpenAI said it has worked with more than 170 mental health professionals to improve how the chatbot responds to sensitive situations. According to expert feedback, the newest GPT-5 model gives more appropriate and consistent replies than older versions.

In testing, GPT-5 met OpenAI’s safety standards in 91% of cases involving suicidal conversations, up from 77% in earlier models. The company also said the model now provides 65% more helpful responses to mental health concerns and performs better in longer conversations, which were previously challenging for safety systems.

OpenAI plans to add new safety checks that focus on emotional dependence and non-suicidal crises to ensure future versions handle such issues responsibly.

Facing Pressure and Legal Scrutiny

These updates come as OpenAI faces increasing scrutiny over the impact of its technology on vulnerable users. The company is currently being sued by parents of a 16-year-old boy who reportedly discussed suicidal thoughts with ChatGPT before his death.

Officials in California and Delaware have also warned OpenAI to strengthen protections for younger users. They suggested possible regulatory actions if the company fails to address these concerns.

OpenAI CEO Sam Altman recently said that the company has made progress in reducing mental health risks on the platform, though he shared no specific evidence. The new report supports some of his claims but also shows how large the problem remains.

Interestingly, while improving safety, OpenAI has also relaxed some content restrictions, such as allowing adult users to have erotic conversations with the chatbot. Older models like GPT-4o, which have fewer safeguards, remain available to millions of subscribers.

Although OpenAI says it is committed to safety, experts note that the long-term effects of AI chatbots on mental health are still unclear. The company’s latest data shows progress, but also highlights how complex and serious these challenges truly are.

Read More: Bitcoin & Ethereum rise as crypto market hits $3.92 trillion