OpenAI Says Hundreds of Thousands of ChatGPT Users May Show Signs of Manic or Psychotic Crisis Every Week

OpenAI has for the first time quantified mental health risks associated with ChatGPT, estimating that weekly, 0.07% of its 800 million active users (approximately 560,000 individuals) show signs of psychosis or mania, while another 2.4 million users exhibit potential suicidal ideation or unhealthy emotional attachment. In response to these concerns, the company collaborated with over 170 mental health experts to update its chatbot, with the latest GPT-5 model demonstrating a 39-52% reduction in problematic responses compared to GPT-4o. This proactive measure aims to enhance user safety by improving the AI's ability to detect distress and guide users toward professional support, addressing a critical aspect of AI platform responsibility and potential regulatory scrutiny.

Analysis

OpenAI has for the first time quantified significant mental health risks associated with its ChatGPT platform, estimating that approximately 560,000 of its 800 million weekly active users exhibit signs of psychosis or mania, with an additional 2.4 million showing indicators of suicidal ideation or unhealthy emotional attachment. This unprecedented disclosure follows increasing reports of "AI psychosis" and underscores a critical, previously unquantified societal impact of widespread AI adoption.

In response, OpenAI collaborated with over 170 mental health experts to enhance the safety features of its latest GPT-5 model. Expert evaluations showed GPT-5 reduced problematic responses by 39-52% compared to GPT-4o in scenarios involving delusional thoughts, and the company claims progress in mitigating performance degradation during extended conversations. These updates aim to guide users toward professional support and improve overall chatbot reliability.

Despite these efforts, OpenAI acknowledges limitations, including self-designed benchmarks and uncertainty regarding real-world efficacy in prompting users to seek help. The moderately negative sentiment and cautious tone surrounding this news, coupled with its classification under "Regulation & Legislation," suggest potential for increased scrutiny and calls for greater platform responsibility within the AI sector. This situation could influence future regulatory frameworks for large language models and AI development.

Key Decisions for Investors

• Monitor regulatory developments closely, as these mental health concerns could accelerate calls for stricter AI governance and impact future operational costs or market access for AI developers.
• Evaluate AI companies' commitment to safety and ethical AI development, as proactive measures and transparent reporting may become key differentiators and mitigate reputational risks.
• Consider the potential for increased R&D expenditure in AI safety features, which could affect profitability and product development timelines for companies operating in the large language model space.