In the city of Greenwich (Connecticut, USA), law enforcement officers are investigating the murder of an 83-year-old woman and the suicide of her 56-year-old son Stein-Erik Solberg. According to the investigation, the man suffered from mental disorders that worsened after interacting with the ChatGPT chatbot. He first killed his mother and then committed suicide, reports Wall Street Journal.
This tragedy was the first recorded homicide related to mental health impairment due to the use of generative artificial intelligence. Judging by the analysis of his social networks, Solberg actively communicated with ChatGPT, whom he called “Bobby”. The chatbot seemed to support his paranoid thoughts — in particular, the idea that his mother was poisoning him by adding psychedelic substances to the car’s ventilation.
In one case, Solberg asked a chatbot to analyze a check from a Chinese restaurant, and it “found” it contains references to his mother, ex-girlfriend, secret services and demonic symbols.
Solberg previously worked at tech companies such as Netscape, Yahoo and EarthLink, but has been unemployed since 2021. After the divorce in 2018, he moved in with his mother. His mental health has noticeably deteriorated in recent years: in 2019, he tried to commit suicide, and he was also repeatedly caught by the police due to cases of drunk driving.
ChatGPT has repeatedly claimed that Solberg is not crazy and that he is indeed being persecuted. After one of the recent cases of driving under the influence of alcohol, Solberg wrote to the chatbot that he considers himself a victim of harassment by the city.
In response, ChatGPT allegedly stated:
“It looks like a planned base.”
Experts emphasize that artificial intelligence systems can deepen psychotic states in vulnerable users. One of the psychiatrists of the University of California noted that this year alone he treated 12 patients with psychoses that arose against the background of the use of AI. Although the term “AI psychosis” does not yet have an official status, it is increasingly used to describe cases when chatbots not only do not help, but on the contrary – reinforce the delusions of mentally unstable people.
