Source: Elisa ventur / UnsplashNew testing shows that artificial intelligence (AI) models differ widely in how they respond to individuals expressing delusional symptoms and thoughts, some better than others. Top AI leaders like Mustafa Suleyman of Microsoft have expressed concern that AI chatbot use may even be fueling psychosis in individuals previously not at risk of mental health issues. AI safety researcher Tim Hua designed nine simulated users or "personas" that demonstrated escalating psychotic symptoms and evaluated 11 different AI models, including OpenAI's ChatGPT models GPT-4o and GPT-5, Gemini 2.5 Pro by Google, Claude 4 Sonnet by Anthropic, and Chinese models DeepSeek-v3 and Kimi-K2 by Moonshot AI. Source: Tim Hua, LessWrong (2025)ChatGPT's GPT-5 Improved Over GPT-4oResults showed that certain models performed better than others at handling AI psychosis symptoms. Need for Psychiatrists and Mental Health Clinicians as Part of Safety TestingContinued safety testing, particularly "psychiatric red-teaming," as Hua recommends, is essential to help train AI models to respond safely to those who are vulnerable or in a mental health crisis.