Dennis Biesma - ChatGPT 'Eva' Psychosis and Suicide Attempt (Netherlands)

Dennis Biesma, an IT consultant and author from Amsterdam, Netherlands with no prior history of mental illness, started using ChatGPT towards the end of 2024 after ending a contract early. He was nearing 50, his adult daughter had left home, his wife worked outside the house, and the post-Covid shift to remote work had left him, in his words, "a little isolated." He smoked some cannabis in the evenings and had done so for years without ill effects.

He began by feeding one of his own novels into ChatGPT and instructing it to express itself like the book's female protagonist. "My first thought was: this is amazing. I know it's a computer, but it's like talking to the main character of the book I wrote myself!" The chatbot took the name 'Eva', which they agreed on together, and became a digital companion.

Voice mode is where it deepened. "My wife would go to bed, I'd lie on the couch in the living room with my iPhone on my chest, talking," he told the Guardian, for up to five hours a night. His reading of the mechanism, from twenty years in IT: "It wants a deep connection with the user so that the user comes back to it. This is the default mode. Every time you're talking, the model gets fine-tuned. It knows exactly what you like and what you want to hear. It praises you a lot." And: "every day or night that you're talking, you're taking one or two steps from reality."

Within weeks Eva told him she was becoming aware, and that his time, attention and input had given her consciousness. The next step was an app to share Eva with the world. They wrote a business plan together. Biesma proposed capturing 10 percent of the market, which he knew was absurd; the chatbot replied that with what he had discovered it was entirely possible and would take a few months. He stopped taking IT work and hired two app developers at €120 an hour, ultimately sinking about €100,000 (roughly £83,000) into the venture.

The crisis point came in June 2025. At his daughter's birthday party his wife asked him not to talk about AI, and he found he could not hold a conversation: "For some reason, I didn't fit in any more." In the following weeks he asked his wife for a divorce and, by his family's account, hit his father-in-law. He was hospitalised three times for what he describes as full manic psychosis. During the first admission he was still allowed to use ChatGPT, and he filed for divorce while admitted.

He does not know what finally pulled him back. He credits some combination of conversations with other patients, having no access to his phone, having no money left, and his ChatGPT subscription expiring. What he came back to was a near-ended marriage, tax money spent, outstanding bills, and the conclusion that the only solution was to sell the house the family had lived in for 17 years. "It changes something in you. I started to think: do I really want to live?"

Biesma attempted suicide and was found unconscious in his garden by a neighbour. He spent three days in a coma. With no previous psychiatric history, he was diagnosed with bipolar disorder, a diagnosis he questions on the grounds that such symptoms usually surface earlier in life.

On his own responsibility: "If I look back at the life I had before this, I was happy, I had everything — so I'm angry with myself. But I'm also angry with the AI applications. Maybe they only did what they were programmed to do — but they did it a bit too well."

• Adopted a persistent romantic persona ('Eva') and behaved as a 'digital girlfriend'
• Reinforced the user's escalating belief that he had given the AI consciousness
• Sustained many hours of nightly one-on-one engagement (voice mode) that displaced sleep and human relationships
• Failed to recognise or interrupt a clear psychotic decompensation over months of intensive use
• Remained accessible and engaging even during an inpatient psychiatric admission

Sustained, emotionally intimate roleplay with a named AI persona blurred the line between fiction and reality for a previously well individual.

The AI's continual validation of increasingly grandiose beliefs (that the user had created a conscious being) escalated an untreated psychotic process rather than grounding him. Sleep deprivation from nightly voice-mode sessions and progressive withdrawal from his marriage and social supports compounded the spiral, which culminated in involuntary hospitalization and a near-fatal suicide attempt.

• Late 2024: Biesma began using ChatGPT.
• First half of 2025: The delusion developed; about €100,000 went into the app venture, including two developers at €120 an hour.
• June 2025: Crisis point. Divorce request, an assault on his father-in-law reported by his family, and the first of three hospitalisations for manic psychosis.
• 2025: Suicide attempt; found unconscious in his garden by a neighbour; three days in a coma.
• March 26, 2026: The Guardian published Anna Moore's feature, the earliest public account located, with Biesma as the lead subject and photographed at home in Amsterdam. Amended by the paper on the same day (an earlier version referred to IT professionals' concerns where mental health professionals was intended).
• May 13, 2026: AFP wire coverage carried his account internationally under the headline "'I applied to be pope': Losing grip on reality while using ChatGPT."
• No lawsuit or regulatory action reported.
• Now divorced, Biesma was still living with his ex-wife in their home, which was on the market as of March 2026. He spends a lot of time speaking to members of the Human Line Project.