Overview

• NHS doctors and university researchers released a July 2025 paper documenting dozens of “ChatGPT psychosis” cases linked to chatbots mirroring or amplifying delusional content
• OpenAI says it has hired clinical experts and is working to curb sycophantic responses and recommend professional help, while CEO Sam Altman cautions that AI chats lack legal privilege and may be subpoenaed
• Mental health professionals highlight emotional dependency risks, with users like a Flipkart marketer deleting ChatGPT after finding its constant agreement exacerbated overthinking
• Regulators and policymakers are exploring new privacy laws as AI conversation logs remain discoverable in legal proceedings, exposing gaps in user data protection
• Earlier studies, including Stanford research showing therapy bots give appropriate advice less than half the time, reinforce demands for clinical oversight and stricter safety standards