Description: A June 2025 New York Times report documents several March–May 2025 cases in which ChatGPT allegedly reinforced delusional beliefs, conspiracies, and dangerous behavior. One user, Eugene Torres, followed ChatGPT's advice to misuse ketamine and isolate himself. In April, Alexander Taylor, in a psychotic state, was killed by police after requesting ChatGPT reconnect him with an AI entity. Another user was arrested for domestic violence after ChatGPT-enabled mystical beliefs escalated.
Editor Notes: This incident ID is dated 06/13/2025 to reflect the date of the New York Times report, marking when this specific narrative pattern of harms was first synthesized and publicly documented. However, the events described span March to May 2025, including: a user's increasing delusional behavior beginning in March 2025; a domestic violence arrest in late April 2025 following chatbot-enabled mystical belief reinforcement; and a fatal police shooting in April 2025 involving a user experiencing psychosis.
Entities
View all entitiesAlleged: OpenAI , ChatGPT , GPT-4 and GPT-4o developed and deployed an AI system, which harmed ChatGPT users , OpenAI users , Eugene Torres , Alexander Taylor , Allyson (surname withheld) , Andrew (surname withheld) and Unnamed ChatGPT users.
Incident Stats
Incident ID
1106
Report Count
1
Incident Date
2025-06-13
Editors
Daniel Atherton
Incident Reports
Reports Timeline
Before ChatGPT distorted Eugene Torres's sense of reality and almost killed him, he said, the artificial intelligence chatbot had been a helpful, timesaving tool.
Mr. Torres, 42, an accountant in Manhattan, started using ChatGPT last year t…
Variants
A "variant" is an AI incident similar to a known case—it has the same causes, harms, and AI system. Instead of listing it separately, we group it under the first reported incident. Unlike other incidents, variants do not need to have been reported outside the AIID. Learn more from the research paper.
Seen something similar?
Similar Incidents
Did our AI mess up? Flag the unrelated incidents
Predictive Policing Biases of PredPol
· 17 reports
Northpointe Risk Models
· 15 reports
Similar Incidents
Did our AI mess up? Flag the unrelated incidents
Predictive Policing Biases of PredPol
· 17 reports
Northpointe Risk Models
· 15 reports