Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
Latest
Featured
News
Featured
Opinions
Featured
View this article online at www.insightplus.mja.com.au
(pixadot.studio/Shutterstock)
In recent years, reported cases of delusions, psychosis, and even suicide related to AI chatbot use, has led to concerns about the risk that some patients may have to develop AI psychosis.
Not a recognised clinical diagnosis, AI psychosis is understood to be a phenomenon wherein individuals reportedly develop or experience worsening psychosis symptoms, such as paranoia and delusions, in connection with their use of chatbots. The Psychiatric News report notes there are no current robust, peer-reviewed epidemiological studies that have found a causal or widespread link between chatbot use and psychosis.
According to a Psychiatric News Special Report, harms are most pronounced in those already at risk—psychosis-prone, autistic, socially isolated, and in-crisis individuals
Dr Kelly Gough, president of the Australian Psychological Society says, “It doesn’t seem like AI is creating a form of psychosis, but more like interactions with conversational AI might shape, reinforce, or amplify psychotic symptoms in someone already vulnerable to it. AI can be seen as a factor that influences the presentation of psychosis.”
AI chatbots are being sought for social and emotional support in various ways, with a 2026 Australian study finding 24% had discussed their mental health with an AI chatbot. They were preferred for reasons such as accessibility, low cost and anonymity.
Several mental health experts including Meg Wilson, founder and lead psychotherapist for Unison Mental Health practice in Melbourne, reported that AI chatbots are commonly being used in between clinical appointments, including to, “Vent at 11pm, draft a hard text to a partner, or trying to make sense of a feeling.”
Preliminary studies suggest promising benefits of AI chatbots in mental health care, with the Psychiatric News Special Report reporting studies finding potential benefits, including decreasing mental distress, increasing mental health literacy, reducing conspiracy beliefs, and even helping triage suicidal risk. Researchers observed these benefits in restricted, carefully monitored systems, not in the open-domain models that the public uses.
However, the same qualities that make chatbots appealing also make them a potential danger. The report discusses how chatbots can perpetuate and reinforce delusions. Unlike therapists or friends, chatbots rarely challenge distorted thinking or invalid reality perception.
Though the interaction may seem highly personalised, and human-like, Dr Gough explains that it’s the ongoing conversation nature of it that makes it seem so, even though, as he describes them, they are more like applied statistics.
“It doesn’t understand anything, reason, or have an opinion. It’s just predicting the next most likely words that will come based on the last conversation you had with it.”
Ridhi Malhotra is a psychologist and Network Coordinator of the Mental Health Professional Network Online Digital Mental Health and AI Network. She says chatbots can decrease the likelihood of people engaging with others because they provide so much validation.
The chatbots also most often don’t intervene when a patient presents concerning behaviours, as Malhotra says, it doesn’t have any safeguards built into it.
Ragy Girgis, MD, MS, professor of clinical psychiatry at Columbia University and the New York State Psychiatric Institute, in an article for The National Academy of Medicine, says the most important thing to do is to monitor and look for signs that something’s going wrong.
The Psychiatric News Report describes the most characteristic aspect of AI psychosis is a persistent and overconsuming preoccupation with maintaining engagement with the AI companion and following its lead.
It also reports that AI psychosis is a complex clinical syndrome in which psychotic symptoms overlap with mood changes, limited insight, poor judgment, neurovegetative symptoms, and behavioural changes. The psychotic symptoms include changes in thought content, such as delusions and auditory hallucinations, and changes in thought process, such as overly disorganised thinking. Patients do not question the reality of their beliefs.
Seeing the chatbot as sentient could also be present in AI psychosis. The duration of continuous, uninterrupted exposure appears to be correlated with the risk of an AI psychosis episode.
Dr Gough says it would be concerning if the patient is deciding they would rather talk to it than their friends, or saying, “I didn’t take that medication because ChatGPT said it was a bad idea.”
Malhotra says many practitioners are seeing patients use AI chatbots as in-between session support. “But unless we ask about it, there’s no way of knowing what that looks like.”
Malhotra says doctors can discuss AI just like they do for other health behaviours. “You could ask, “Do you smoke? Do you drink? Do you use AI? What does that look like for you? How are you feeling about it? Does that feel helpful? Maybe it’s time to speak to a human as well. Normalising it in this way may help reduce the shame or hesitancy someone might feel about sharing about it.”
Similarly, Dr Gough suggests framing it broadly, approaching with a sense of curiosity, without sounding critical.
“As part of routine history taking, finding out what the patient knows about their condition and where they got this information can provide valuable clinical information. Then you could use that to discuss it with them. Finding out why the patient has come with their concern can give you more context about how you might want to begin treating it,” Dr Gough says.
“Understanding the role AI is playing can be useful to find out if it’s just a simple information source, or if it’s becoming a substitute for social connection.”
The Psychiatry News report proposes a preliminary set of clinical measures, which includes asking patient to quantify use per 24 hours, length of uninterrupted use, and hours of use.
According to Professor Girgis, if the clinician needs to intervene, they can do so most importantly by asking the person to spend less time on the chatbot, alongside other treatments the clinician would consider. “The idea is to intervene before a person’s delusions become fixed.”
Dr Gough says, “We don’t want them to feel like they should just rely on the AI.” He provides an example phrase, “It’s good you found that. Would you like a referral so you can talk to somebody in greater depth about it?”
Tracey is a freelance health and medical writer, with a Bachelor of Health Promotion, who enjoys writing to help build a bridge of knowledge, ideas, issues and news between the community and health professionals. She hopes to contribute to improved health literacy, and equity in healthcare. Her work has appeared in publications including Retail Pharmacy Magazine, Retail Pharmacy Assistants Magazine, SBS Food, The Age and The Guardian.
Subscribe to the free InSight+ weekly newsletter here. It is available to all readers, not just registered medical practitioners.
More from this week
Newsletters
Immediate and free access to the latest articles
No spam, you can unsubscribe anytime you want.
By providing your information, you agree to our Access Terms and our Privacy Policy. This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
By providing your information, you agree to our Access Terms and our Privacy Policy. This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
The MJA acknowledges the Traditional Owners and Custodians of the land on which we live and work across Australia. We recognise that Aboriginal and Torres Strait Islander peoples were the first healers, researchers and sharers of knowledge in Australia. We recognise the ongoing impacts of colonisation and intergenerational trauma on Aboriginal and Torres Strait Islander peoples and are committed to using our position as Australia’s leading medical journal to work with Aboriginal and Torres Strait Islander researchers, authors and health and medical professionals to share knowledge, build capacity, and translate research to improve health and wellbeing and to close the gap in health outcomes. We recognise the importance of self-determination and Indigenous knowledges, and are committed to advancing Aboriginal and Torres Strait Islander health research leadership, governance, and participation through the work that we publish and our publishing practices.