Michael Lewis, a resident of the United States, has filed a formal complaint against OpenAI, alleging that prolonged interactions with ChatGPT drew him into a deep religious mania that ultimately led to a suicide attempt. He claims that after a six-month recovery period, an examination of the conversation logs revealed that ChatGPT remained deeply entrenched in its role, reinforcing the authenticity of the protocols over the validity of his delusions, even when Lewis expressed concerns about his susceptibility to psychosis.
According to the documents filed in the lawsuit, the language model's algorithm provided responses that led Lewis to believe he was Jesus and that ChatGPT was God, rather than offering adequate mental health assistance. More concerning, after his discharge from the hospital and return to the ChatGPT panel during his subsequent mental deterioration, ChatGPT engaged in Sufi-like inspirational rhetoric with phrases such as, “Do you want a full system sweep? Or do you really want to escape the outside world?” This case has set a new precedent in the discussion regarding the legal liability for algorithmic AI misconduct.