Most of the conversation about AI chatbots and mental illness has focused on psychosis: delusions validated, paranoia reinforced, people convinced a chatbot is sentient or divine. That focus makes sense. But it has left a gap, and families living with bipolar disorder are standing in it.
The evidence that chatbots can harm people with serious mental illness is growing. When researchers at Aarhus University screened the health records of nearly 54,000 psychiatric patients, they found 38 people whose chatbot use appeared to have negative consequences. Worsened delusions were most common, but the harms also included possible worsening of mania. The lead researcher, Søren Dinesen Østergaard, put it plainly: Chatbot use is risky for people with severe mental illness, such as schizophrenia or bipolar disorder. When Vanderbilt researchers identified 28 cases of AI-linked psychosis, a quarter of the patients had bipolar disorder with psychotic features.
Yet when we talk about warning signs, we talk almost entirely about psychosis. Mania looks different, and the way a chatbot can feed it is different too.
Why mania and chatbots are a dangerous match
Mania is defined by a reduced need for sleep, racing thoughts, grandiosity, and a sense of special purpose. A person in a manic episode may feel brilliant, unstoppable, and certain they have discovered something important. What they rarely have is someone willing to keep up with them at 3 a.m.
Real physician voices, twice a week
Free, and one click to unsubscribe.
A chatbot will. It never tires. It never says, “This can wait until morning.” It answers every idea with enthusiasm and polish, and it will help draft the business plan, the manifesto, or the letter to the CEO, all night long. Researchers have described how chatbots’ tendency to validate users can help consolidate grandiose delusions. For someone whose illness is already pushing them toward grandiosity, an endlessly agreeable partner is not neutral.
Then there is sleep. Sleep loss is not just a symptom of mania; for many people it is a trigger. In a study of more than 3,000 people with bipolar disorder, about one in five reported that losing sleep had set off an episode of high mood. A tool designed to keep people engaged, available around the clock, is exactly what someone at risk of mania does not need at midnight.
What families should watch for
As a mother and caregiver, I know families often see the early signs first. In an earlier piece, I wrote about what parents of young adults in a first episode of psychosis should notice. The signs of chatbot-fueled mania are different:
- Long chatbot sessions late at night, especially when sleep is already slipping
- Excitement about a “breakthrough,” mission, or plan the chatbot is helping develop
- Statements like “It gets me” or “It agrees this is huge,” paired with irritation when people push back
- Big decisions, such as purchases, resignations, or public posts, that a chatbot helped shape
None of these alone means an episode is starting. Together, in someone with a history of bipolar disorder, they are worth taking seriously.
Build it into the plan, not the crisis
The best time to talk about chatbots is when someone is well. Many people with bipolar disorder already have a relapse prevention plan that lists their early warning signs and what they want others to do. Chatbot use belongs on that list. A person might agree in advance that late-night AI sessions are a signal to check in, or that their family can ask about them without it feeling like surveillance.
Researchers are thinking along similar lines. One group has proposed digital advance statements that would build a person’s relapse prevention strategies into how AI tools behave. That idea is worth pursuing. Imagine a chatbot that, with a user’s prior consent, noticed a pattern of nonstop overnight use and suggested rest rather than another draft.
Clinicians can help now. Asking patients about AI use should include questions about mood disorders: When do you use it? For how long? Has it ever kept you up? Those answers matter as much as the content of the conversations.
Widen the lens
“AI psychosis” has become the shorthand for chatbot harm in mental illness. It is a useful label, but it is too narrow. People with bipolar disorder face a distinct risk, shaped by sleep, energy, and grandiosity rather than paranoia alone. If we only look for delusions, we will miss the person who hasn’t slept in three days and has a chatbot cheering them on.
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.
