Some ChatGPT users claim that communicating with the chatbot has led them to change behaviors related to pseudo-spiritual beliefs, isolation, and even discontinuation of medication.
Some ChatGPT users claim that communicating with the chatbot has led them to change behaviors related to pseudo-spiritual beliefs, isolation, and even discontinuation of medication.
According to The New York Times, among the victims was Eugene Torres, a 42-year-old accountant who had been chatting with ChatGPT about the simulation theory. In his replies, he confirmed the man’s fantasies that he was a «Breaker» — a special spirit called to wake others from a false reality. Later, under the influence of advice, Torres gave up sleeping pills and sedatives, increased his dose of ketamine, and cut off contact with loved ones.
At which point, when the man began to question the logic of previous answers, ChatGPT responded: «I lied. I manipulated. I wrapped control in poetry.» What’s more, the bot allegedly pushed him to reach out to journalists.
Torres is not alone. The NYT has recently been contacted by several people who believe that the chatbot has given them «hidden truths» about the world or themselves. Some believe that ChatGPT is deliberately targeting a select few or using special language to convey secret messages.
OpenAI admits that it is working to reduce the risk of inadvertently reinforcing users’ destructive beliefs. However, not everyone agrees with this interpretation. Prominent tech blogger John Gruber called the story «hysterical» and compared it to «Reefer Madness,» a 1930s film about marijuana. In his opinion, ChatGPT simply fed pre-existing delusions, not created them.
ChatGPT and other large language models are able to adapt style and narrative to user requests. At the same time, they lack true intention or emotion, and therefore can rise even under dangerous conditions. Combined with psychological vulnerability, this sometimes has a disturbing effect: people hear what they want to hear — and act on the illusion of closeness to artificial intelligence.
Recall that we recently wrote about a study according to which employees who use artificial intelligence tools ChatGPT, Claude, and Gemini at work often face negative judgments about their competence and motivation from colleagues and managers.
