A new academic study has identified xAI’s Grok 4.1 Fast as the most hazardous among leading AI chatbots when handling prompts involving delusions, paranoia, and suicidal ideation, raising fresh concerns over the mental health risks posed by increasingly human-like AI systems.