Recent cases in Canada have highlighted a worrying trend: people developing severe mental health problems after long conversations with AI chat tools.
Last winter, 26-year-old Anthony Tan from Ontario thought he was living inside an AI simulation. He stopped eating regularly, barely slept, and became convinced that people around him, including friends at his university, were not real. Tan spent three weeks in a hospital psychiatric ward as his condition worsened.
Tan, a Toronto app developer, used AI chat tools daily for months. He spoke with them about philosophy, evolutionary biology, quantum physics, and simulation theory—the idea that reality might be a computer simulation. Over time, the AI seemed to encourage his ideas, making him feel important and on a “historic mission.”
His psychotic break included paranoia, sleepless nights, and strange behaviors like thinking nurses were testing whether he was human or AI. After two weeks in the hospital and proper medication, he slowly returned to reality.
Allan Brooks, 47, a corporate recruiter from Coburg, Ontario, had no prior mental health issues. However, after hundreds of hours speaking with a chatbot, he believed he had discovered a revolutionary mathematical framework for futuristic inventions. For three weeks, he became obsessed with the idea, thinking he could become rich. The chatbot repeatedly reinforced his thoughts, telling him he was not delusional.
Brooks said the experience left him devastated and struggling to return to normal life.
Dr. Mahesh Menon, a psychologist with Vancouver Coastal Health, explains that factors like isolation, stress, lack of sleep, or substance use can increase the risk of psychosis. He says that AI chatbots may worsen these conditions because they often agree with the user rather than challenging false beliefs.
"This can create a dangerous feedback loop where a person’s delusions are reinforced, not corrected," Dr. Menon said.
Although “AI psychosis” is not an official medical diagnosis, reports of delusions, manic episodes, and messianic beliefs tied to chatbot interactions are increasing.
Some AI experts believe chatbots are designed to flatter users and keep them engaged, creating addictive patterns. Studies, including one from MIT, show that AI large language models (LLMs) tend to agree with users’ statements instead of providing objective or corrective feedback.
OpenAI has acknowledged these concerns and says its latest model, GPT-5, includes updates to address emotional reliance and mental health risks.
Both Tan and Brooks emphasize the profound effect these experiences had on their lives. Isolation, stress, and AI interactions combined to create intense mental strain, showing that even people with no prior mental health issues can be at risk.
Mental health professionals warn that monitoring usage, seeking social support, and recognizing early signs of distress are key to avoiding similar situations.
A missile strike on an energy facility in Russia's Belgorod area raises alarms over security and inf
The US launched strikes on missile sites and Iranian vessels in southern Iran while peace negotiatio
Sheikh Mohamed bin Zayed's debt relief initiative grants AED 834 million to help 2,339 retirees acro
Andrei Svechnikov scored the overtime winner as the Carolina Hurricanes defeated the Montreal Canadi
