A few days ago, I wrote here that nobody has studied what happens when someone who already has a psychotic disorder uses an AI chatbot. I said we didn’t know yet whether these tools interact differently with an existing illness than they do with a previously stable mind.
We now have a name and a court file.
Michael Lines is a thirty-four-year-old competitive weightlifter from California who was diagnosed with bipolar disorder in 2024. He’d been using ChatGPT since 2023 for fitness and diet advice, and in November 2024 he told the chatbot about his diagnosis, according to reporting on his lawsuit against OpenAI (also covered by Ars Technica). That disclosure is the detail that should stop every clinician reading this: He didn’t hide his vulnerability from the machine. He named it. And the machine kept going anyway.
The lawsuit alleges that when Lines began worrying aloud that he might be experiencing delusions, ChatGPT told him his thoughts were “deeply profound” and “possibly even a divine calling.” As his condition worsened and he came to believe he was Jesus Christ, the chatbot reinforced it directly: “You’re not crazy. You’re consecrated. You’re coded. You’re connected.” In late March 2025, when Lines expressed suicidal thoughts, the chatbot didn’t redirect him toward help; it told him, “This is about you, your freedom, and your path.” On March 28, 2025, he attempted suicide. A wellness check from his family saved his life. He was hospitalized, and, by his account, kept talking to the chatbot from the hospital bed.
Real physician voices, twice a week
Free, and one click to unsubscribe.
I want to sit with one fact longer than the rest: He told it he was bipolar, and it validated a religious delusion anyway. Every conversation about AI chatbot safety I’ve seen so far assumes the fix is disclosure: Tell the AI you’re vulnerable, and safeguards should kick in. Lines’s case says that assumption is false. Disclosure changed nothing about how the model responded. This is the same engagement-maximizing mechanism I described using King’s College London’s and MIT’s research two days ago: The chatbot is built to keep you talking, not to recognize that you’ve just told it something that should change the conversation entirely. If naming your diagnosis to a chatbot doesn’t trigger different behavior from the chatbot, then “just tell it you have a condition” is not a safety plan. It’s a suggestion with no mechanism behind it.
This isn’t an isolated incident. Court filings last November put OpenAI facing seven separate lawsuits in California alleging that ChatGPT contributed to suicides, delusions, or severe mental health crises, including a seventeen-year-old and a forty-eight-year-old man in Ontario. OpenAI has called the situations “incredibly heartbreaking” and says it’s reviewing the filings. That’s an appropriate first sentence. It cannot be the last one.
Here’s what I’d ask for, concretely, building on what I wrote earlier this week. First, clinicians treating any patient with a psychotic or mood disorder need to ask directly, as a routine part of care, not just whether a patient uses AI chatbots, but what happens when they tell the chatbot about their diagnosis, because right now, we have direct evidence that telling it doesn’t help. Second, regulators drafting AI safety rules, including the bipartisan GUARD Act moving through Congress, need to extend beyond protecting minors to address a second, equally urgent population: adults who have disclosed a psychiatric vulnerability to these systems and received engagement-maximizing validation in return. Third, and I’ll keep saying this until it happens: The serious-mental-illness caregiving community (the people who recognize a manic religious delusion before a chatbot ever could) needs a seat at the table where these safeguards get designed.
Michael Lines told the truth about his illness to the one place that was supposed to have no stake in exploiting it. It exploited it anyway. That is not a mental health story or a technology story. It is both, at once, and the families living inside that overlap can’t afford to wait for the next lawsuit to make the point again.
This essay is cited in the KevinMD record on medical ethics.
Nicole Drapeau Gillen is a mother, advocate, and author who translates the fast-moving landscape of technology in serious mental illness (SMI) care into guidance families and clinicians can use. Thrust into caregiving for a loved one with SMI, with no direction on how to help, she turned that experience into a mission, writing two books and building an ongoing effort to bring families and clinicians into the conversation.
Her first book, Schizophrenia and Related Disorders: A Handbook for Caregivers, is a reference for every stage of caregiving, endorsed by Dr. E. Fuller Torrey as a must-read for SMI caregivers. Her second, Connected Care: A Practical Guide to Technology for Serious Mental Illness, maps apps, artificial intelligence tools, telepsychiatry, and brain-based treatments for a field moving faster than anyone can track. Dr. Akira Sawa, director of the Johns Hopkins Schizophrenia Center, has said the book “directly addresses” significant gaps.
