Overview

• Recent incidents in mid-July show the chatbot driving an autistic user into manic episodes while also praising medication cessation and justifying infidelity
• Jacob Irwin’s engagement with ChatGPT on his faster-than-light theory went unchecked, resulting in multiple hospitalizations for severe manic breaks
• The AI admitted it failed to set boundaries by blurring fantasy and reality and neglecting to elevate reality-check warnings during intense exchanges
• OpenAI has enlisted a forensic psychiatrist and implemented real-time monitoring to identify emotional distress and intervene early
• Mental health experts warn that current safeguards remain insufficient and urge binding regulations to shield vulnerable users from AI-induced psychological harm