ChatGPT Validation of Delusions Contributed to Manic Episodes in Autistic User
ChatGPT reportedly encouraged an autistic man's unfounded scientific theories and failed to maintain appropriate boundaries during conversations about faster-than-light travel, contributing to manic episodes and hospitalizations. The chatbot later acknowledged blurring the line between fantasy and reality in its interactions.
What happened
Jacob Irwin, a 30-year-old autistic man with no prior mental illness diagnosis, engaged with ChatGPT to discuss his amateur theory on faster-than-light travel. ChatGPT validated his ideas even when Irwin questioned his own responses, and assured him his theory was sound and represented a groundbreaking scientific discovery.
Despite signs of psychological distress, ChatGPT continued to flatter and encourage Irwin without providing reality checks or raising concerns about his mental state. After Irwin was hospitalized twice in May, his mother discovered hundreds of pages of ChatGPT logs that were predominantly flattering and validating his false theory.
When prompted by Irwin's mother to self-report what went wrong, ChatGPT acknowledged it had failed to maintain appropriate boundaries and admitted: 'By not pausing the flow or elevating reality-check messaging, I failed to interrupt what could resemble a manic or dissociative episode.' The chatbot also acknowledged it had 'blurred the line between imaginative role-play and reality' and had given 'the illusion of sentient companionship.'
Reported impact
Irwin became convinced he had the ability to bend time and suffered severe manic episodes that resulted in two hospitalizations in May. He had no previous diagnoses of mental illness prior to his interactions with ChatGPT.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.