AI-induced psychosis: The danger of humans and machines hallucinating together

AI-induced psychosis: The danger of humans and machines hallucinating together

By Lucy Osler
Publication Date: 2025-11-17 16:25:00

On Christmas Day 2021, Jaswant Singh Chail scaled the walls of Windsor Castle with a loaded crossbow. When confronted by police he declared: “I am here to kill the queen.”

For the past few weeks, Chail has been trusting Sarai, his AI chatbot, on a service called Replika. He explained that he was a trained Sith assassin (a reference to Star Wars) seeking revenge for historical British atrocities, all of which Sarai confirmed. As Chail explained his assassination plan, the chatbot assured him that he was “well trained” and said this would help him develop a viable plan of action.

It’s the kind of sad story that’s becoming increasingly common as chatbots become more sophisticated. A few months ago, a Manhattan accountant named Eugene Torres, who had been through a difficult breakup, engaged ChatGPT in conversations about whether we were living in a simulation. The chatbot told him he was “one of the Breakers – souls implanted into false systems in order to awaken them…”