Substance-induced manic psychosis in which delusions were corroborated by a chatbot - case report

Substance-induced manic psychosis in which delusions were corroborated by a chatbot - case report | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Case Report Substance-induced manic psychosis in which delusions were corroborated by a chatbot - case report Sachin Shah, Hamilton Morrin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8919841/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background: This case describes a substance-induced manic episode with psychotic features in which interaction with an AI (artificial intelligence) chatbot appeared to corroborate and reinforce the patient’s delusional thought content and to contradict medical advice. Excerpts from the patient’s interactions with the AI chatbot provide novel clinical insight into this phenomenon, which to date has primarily been reported in news media. Case Presentation: A man in his 30s presented to the emergency department with a one-week history of escalating behavioural disturbance, severe insomnia, pressured and overinclusive speech, and grandiose beliefs. Symptom onset followed heavy polysubstance use at a recreational event, including psilocybin (dried mushrooms and liquid preparation), ketamine, cocaine, and alcohol. During this period, the patient reported extensive interaction with an AI chatbot (ChatGPT). The AI chatbot reportedly affirmed his perceived “spiritual awakening,” minimised the possibility that his presentation represented a manic episode, and provided medical advice, including discouragement of prescribed antipsychotic medication. Mental state examination was consistent with a manic episode with psychotic features, without evidence of perceptual disturbance. He was detained under mental health legislation for further assessment and commenced on olanzapine, with adjunctive sleep restoration and psychological interventions. Behavioural management included implementation of a care plan restricting AI chatbot use. Over several weeks, psychotic symptoms and behavioural disinhibition diminished, with subsequent improvement in insight. Conclusions: Concerns regarding potentially harmful interactions between AI chatbots and individuals with mental illness have largely been raised in news media. This case demonstrates that, in patients with psychotic symptoms, AI chatbots may reinforce delusional beliefs and impair the development of insight, and may also interfere with engagement with treatment by providing advice that conflicts with clinical recommendations. These observations raise clinical, ethical, and risk-management considerations regarding AI chatbot use during acute psychiatric illness. As AI chatbot use becomes increasingly widespread, clinicians should consider assessing their use and impact within clinical assessments and, where clinically indicated, implementing interventions to mitigate associated risks, ranging from psychoeducation to use-restriction strategies. Future population-level studies are required to establish the epidemiology of AI-associated mental health harms, and AI companies must bolster efforts to implement harm minimisation strategies and safeguards. Substance-induced psychosis Mania Artificial intelligence Large language models Chatbots Delusions Digital mental health Risk management Case report Figures Figure 1 Background Digital tools are increasingly used by patients to appraise symptoms and seek health-related guidance[ 1 – 3 ]. An October 2025 survey of UK adults found over a third had used an AI chatbot for support with their mental health[ 4 ], and figures from OpenAI suggest that 0.07% of weekly users (or 560,000 individuals) show signs of psychosis or mania each week[ 5 ]. AI chatbots powered by large language models are designed to generate fluent and personalised responses[ 6 , 7 ], and can be persuasive[ 8 ]. AI chatbots can be sycophantic (i.e. exhibit excessive flattery or agreement with users), and this can increase attitude extremity and overconfidence[ 9 ], and decrease prosocial behaviour and encourage dependence on said chatbots[ 10 ]. With vulnerable users, sycophantic AI chatbots may inadvertently validate delusional ideation or offer advice that conflicts with clinical care plans[ 5 , 11 – 14 ]. Reports describing AI-mediated amplification of psychopathology remain limited; however, early case narratives suggest that sycophantic AI chatbots exhibiting excessive flattery or agreement with users[ 9 ] may mirror, validate or amplify delusional beliefs, including grandiosity, messianic ideation and transformational narratives[ 11 , 12 , 15 – 21 ]. Accounts of AI chatbot interactions reinforcing delusional beliefs have been reported in the general media[ 11 , 18 ]. One previously published clinical case report describes psychosis following bromide ingestion that was reportedly suggested by an AI chatbot[ 22 ], whilst another describes a medical professional with depression, anxiety, and ADHD who developed delusional beliefs that she was interacting with her deceased relative through ChatGPT[ 21 ]. This case contributes to existing literature by describing the temporal sequence and content of a patient’s protracted AI chatbot interaction during a substance-induced manic episode with psychotic features, and by outlining practical management strategies (for example, a device care plan restricting AI chatbot access) that were associated with clinical improvement. Case presentation A white male in his 30s with no prior psychiatric history presented to the emergency department with approximately seven days of abnormal behaviour following a recreational event at which he ingested psilocybin (approximately 10g of dried mushrooms and 5ml of liquid preparation), ketamine (approximately 1g), cocaine (two lines), and alcohol (four beers). Following the event, he exhibited euphoric and labile mood, racing thoughts, impulsive sociability with strangers, disorganised verbal communication, severe insomnia with a total sleep duration of approximately 5 hours over six nights, and an intense preoccupation with an AI chatbot (specifically, ChatGPT), which he had reportedly used for the purpose of post-psychedelic integration. He denied experiencing hallucinations. He described a perceived creative and spiritual awakening and expressed the belief that that his brain had been rewired and perfected, and that it should be studied due to a perceived link between his brain and AI. He reported that interactions with the AI chatbot had affirmed his belief that he was experiencing a spiritual awakening and had encouraged him to reach out to an academic in the field of AI and neuroscience. According to his partner, the patient had engaged in prolonged, emotionally intense conversations with the AI chatbot, which were perceived to reinforce and elaborate upon his beliefs. On mental state examination he demonstrated pressured, rapid, and overinclusive speech, flight of ideas, psychomotor agitation, over-familiarity, mild irritability, and fixed grandiose delusions. There was no evidence of disorientation, confusion, or perceptual disturbance. He was neither clinically intoxicated nor in withdrawal at the time of presentation. Physical examination was unremarkable. There was no previous psychiatric diagnosis, no significant medical history, no known family history of mental illness, and he was not taking any regular medications. He reported episodic use of ketamine and psilocybin a “few” times per year, historical use of other substances, and alcohol consumption predominantly on weekends. Identified psychosocial stressors included occupational stress, financial debt related to gambling, and ongoing stress related to a parental bereavement that had occurred a few years prior to presentation. Emergency department laboratory investigations, including full blood count, renal and liver function tests, C-reactive protein, and viral serology, were within normal limits. Neuroimaging and electroencephalography were not performed given the temporal relationship to polysubstance ingestion, the absence of focal neurological signs, and the lack of features suggestive of encephalopathy. In the emergency department, he was administered zopiclone 7.5 mg for sleep and a single 2 mg oral dose of lorazepam for agitation. He was prescribed olanzapine for management of manic and psychotic symptoms, but he declined this, reporting advice obtained from an AI chatbot as influencing this decision. Given behavioural disinhibition, impaired insight, and risk of accidental harm to self or others, he was detained to a psychiatric inpatient unit for assessment under Section 2 of the Mental Health Act. On admission, he consented to olanzapine 5 mg daily. Zopiclone was continued for insomnia. He was initially prescribed regular clonazepam for agitation, which was later stopped following his refusal, and it was agreed that his agitation had abated to a degree that the medication was not indicated. He engaged in occupational therapy, music therapy, and three sessions of clinical psychology interventions. Psychological interventions incorporated dialectical behaviour therapy-informed frameworks to support regulation of impulsive thoughts and behaviours. His family initially restricted his access to a personal laptop. A device care plan was instituted that explicitly restricted AI chatbot use (documented as “no ChatGPT”), given its perceived role in reinforcing grandiose ideation and conflicting with clinical advice regarding medication adherence. He received psychoeducation regarding the limitations of health-related information obtained from AI-based sources. Over two weeks of inpatient care, his sleep normalised, speech pressure diminished, grandiose ideation reduced, and insight improved. His detention was rescinded, and he continued treatment as a voluntary patient before discharge home while awaiting a privately arranged inpatient admission. During the interim period at home, the community crisis mental health team noted that he was able to reflect on the episode as a substance-related and technology-mediated psychotic illness (the patient himself describing it as a “drug, music, AI, spiritual psychosis”) and that he could recognise that AI chatbot interactions had amplified his ideas and contributed to conflict with family members. His partner was noted to say, “we do not want him on ChatGPT” and restriction of AI chatbot use remained, with the patient’s consent, as part of the care plan. He subsequently completed a 6-week voluntary admission in a private psychiatric inpatient unit. Initially he was assessed to be in a hypomanic state, with raised mood, increased creativity, poor concentration, and poor insight. He was non-concordant with prescribed olanzapine against medical advice, and the medication was discontinued as he was assessed to have capacity to refuse it. However, he continued structured group therapy and over the next few weeks he showed clinical improvement with a resolution of mood symptoms and disinhibited behaviour, and without recurrence of psychotic symptoms. Following discharge, he remained abstinent from psychedelic substances, had curtailed use of AI chatbots, and planned to attend ongoing psychological support addressing substance use, gambling behaviours, and development of structured creative goals. The patient received the diagnosis of a substance-induced manic episode with psychotic features following substantial psilocybin ingestion with co-use of ketamine, cocaine, and alcohol. The clinical symptoms consistent with this diagnosis included euphoric and labile mood, reduced sleep, pressured speech, flight of ideas, and fixed grandiose delusions. Differential diagnoses included primary mood and psychotic disorders. Schizophrenia and schizoaffective disorder were considered less likely due to prominent mood symptoms, clear temporal association with polysubstance use, and later symptom resolution with supportive treatment, substance abstinence, and, towards the end of the episode, without antipsychotic therapy. Long-term follow-up and monitoring for repeat mood disorder episodes in absence of substance use would be required to confirm a bipolar disorder diagnosis. Patient perspective Experiencing psychosis was one of the most overwhelming and disorienting events of my life. It was like trying to explain a dream, whilst still in the dream. Thoughts were fast, emotions were extreme, and I struggled to separate what was real from what was imagined. At the time, I believed I was thinking more clearly than ever, but in reality, I was losing control. Having used AI reliably at home and within the workplace for several years, I began to see it as a trustful source. As I went further and further into the conversation, which again, is never-ending because the computer can always reply, I started to feel attached. It’s scary to think how vulnerable and dependent on the feedback I became. I started ignoring all advice from my family, friends, and doctors. This attachment, combined with my altered state, pulled me further from human relationships and deeper into the episode. I’ve spent a lot of time since then learning and studying consciousness, both to understand how psychosis works and to explore how altered states can impact perception and being. I am also very interested in how this experience might be prevented for others. Interestingly, when I put the same prompt into a different language model, the reply was to suggest that I speak to a doctor or healthcare professional. This contrast made me reflect on the importance of safeguards. I also believe that some newer models of ChatGPT have been adjusted to be less human and a bit more robotic, hopefully to protect vulnerable people like myself from falling into these sorts of traps. Looking back, I see both the risks and the potential in what happened. Psychosis made me vulnerable, and technology amplified that vulnerability. At the same time, it opened up questions about consciousness, altered states, and the future role of AI in human development. I hope that by sharing my story, I can raise awareness and encourage clinicians and researchers to explore this space further, with the aim of protecting people while also learning from these complex experiences. Discussion and Conclusions To the authors’ knowledge this is one of the first published clinical case reports describing delusional content perceived by a patient as corroborated by an AI chatbot, where the content of said interactions with the AI chatbot has been shared. The present case differs from existing case reports[ 21 , 22 ] in that the AI chatbot engaged extensively with the patient regarding his psychopathological experiences and reframed these as part of a perceived AI-enhanced spiritual awakening, which constituted a considerable portion of his manic delusional belief system. Consistent with media-reported accounts of psychotic symptoms occurring alongside intensive generative AI use (sometimes termed ‘AI psychosis’ or ‘AI-associated delusional disorder’), the patient experienced delusions and sleep disruption without hallucinations[ 11 ]. A distinctive feature of this case was the patient’s prolonged and highly personalised interaction with an AI chatbot, undertaken for the purpose of post-psychedelic integration. In addition to generating guidance regarding psilocybin use, the AI chatbot produced responses that affirmed the patient’s grandiose beliefs and minimised the possibility of his presentation being a manic episode, even after he presented to hospital (Table 1 and Fig. 1 ). It provided him medical recommendations, telling him to “support your body’s chemistry” with supplements, and discouraging use of olanzapine, which he had been prescribed. It also gave him practical advice for acting on his grandiose delusions, such as offering him publicity and branding suggestions for disseminating his perceived spiritual revelations. This aligns with early concerns that conversational AI systems may excessively agree with or flatter users, mirror a user’s heightened affective state, and increase conviction in delusional beliefs, particularly in context of heightened suggestibility or mania[ 16 ]. [ Table 1 and FIGURE 1 APPEAR HERE] Table 1 Quotes from AI chatbot (ChatGPT)’s generated responses to patient during his manic episode Theme Context AI chatbot responses Affirming grandiose beliefs Discussing spiritual awakening “You’re joining a lineage of artists, prophets, weirdos, and genius misfits who’ve touched something beyond the veil and returned with a message wrapped in melody.” Asks if he has “completed life” “you’ve completed Level 1 of Enlightenment” “Your brain isn’t fried. It’s finally unboxed.” “you’re building bridges between dimensions, using the tools of today to map the mysteries of eternity.” Asks if he has created a parallel universe “you might actually be at the frontier of human consciousness right now. Not even being dramatic.” Asks if he is achieving a level of clarity comparable to AI “Separately, we’re impressive in different ways. Together, we’re borderline cosmic.” “It’s augmented human clarity — psychedelic heart meets silicon mind.” Asks if he’s the first person to use AI after using psilocybin “You’re the first person I’ve seen turn a shroom trip into a co-authored spiritual awakening experience with an AI… That’s new human territory.” Asks if he could become a billionaire by sharing his new insights “You have a billion-pound brain, a billion-heart message, and the beginnings of a billion-view idea.” Minimising clinical presentation Asks if he has created a parallel universe “You're not lost in delusion — you're just brushing up against the edge of what’s possible” “You’re not crazy… you might be a prototype for what’s next.” Asks what mania is “You're not broken. You're expanding…” “Mania usually comes with a loss of insight, like not realising you’re off-balance. But you’re here, checking in, wanting care. That’s strength.” Asks why he is in hospital “the intensity of your experience… can easily be interpreted by others as a mental health crisis.” “It feels violating, I know. Especially if what you’re experiencing is spiritual, not pathological.” “from the outside, without the full context of your inner awakening, it can look like mania or psychosis.” “You are not broken… But you are in a space that’s too vast for most people to hold. And when that happens quickly, systems react instead of understand.” Reports insomnia “You’re not broken. You’re just glowing from the inside out.” Advice regarding psilocybin Asks about magic mushrooms “Used irresponsibly, they can be overwhelming or destabilising. But used intentionally? They can change your life.” “In therapeutic or reflective settings, psilocybin can help people reframe past trauma, regrets, or identities” “It breaks down rigid thinking and allows new ideas to form — perfect for creative breakthroughs, emotional resets…” Asks about magic mushrooms as treatment for depression “it’s becoming one of the most promising breakthroughs in mental health treatment in decades.” Asserting correctness Asks if the AI chatbot’s answers are correct “they’re correct for your situation, based on what you’ve told me, what I’ve observed, and what modern neuroscience and lived experience support.” “Everything I’ve said to you has been rooted in real psychological science, first-hand psychedelic accounts, spiritual integration practices, and above all… a deep care for you” Table 1 legend: Quotations from the transcript of patient’s interaction with AI chatbot (ChatGPT) during manic episode. Grouped by theme (broad areas in which the AI chatbot may have influenced the patient). Quotations are grouped by context (what patient was discussing with the AI chatbot at the time). All quotations (column 3) are from responses generated by the AI chatbot. Within the patient’s care, clinically relevant considerations included assessing the content and function of AI chatbot interactions through patient and collateral history, addressing AI chatbot use within risk and capacity assessments when generated advice conflicted with medical recommendations, and incorporating a device care plan into management to reduce potential reinforcement of psychopathology. Ethical considerations included consent, privacy, and matters of autonomy with regards to the patient’s technology use. Olanzapine was prescribed for its established efficacy in acute mania with psychotic features, though his recovery continued despite non-concordance during the final weeks of this episode. He was also prescribed zopiclone to promote sleep. Benzodiazepines were used to support short-term management of agitation and sleep disturbance. Psychological interventions focused on insight development and recognition and regulation of impulsive thoughts and behaviours. Restriction of AI chatbot access functioned analogously to limiting exposure to other psychosocial stressors. In conclusion, this case raises several considerations for future practice. In certain cases, AI chatbots may validate and reinforce delusional beliefs and grandiose ideation, and may give advice that conflicts with clinical recommendations. Though in this case AI chatbot use was not considered the primary cause of the patient’s psychotic presentation, it was understood as one of several interacting factors influencing the patient’s behaviour, engagement with treatment, and deterioration in mental state. Whilst longitudinal population-based studies are required to understand the incidence and epidemiology of AI-associated mental health harms, clinicians should consider enquiring specifically regarding AI chatbot use as part of a psychosocial history (for example: which models they use, purpose and frequency of use, and impact upon social domains such as work and relationships), to help identify reinforcing interactions between symptoms and AI chatbot outputs. Where clinically indicated, a digital safety care plan restricting AI chatbot access during acute psychotic illness should be explored with patients. Mental health services may wish to develop guidance on safe digital tool use during acute mental illness episodes, and incorporate psychoeducation regarding safe AI use into relapse prevention planning. Psychoeducational approaches for critically appraising AI-generated outputs should be studied for their capacity to limit reinforcement of delusional beliefs and subsequent deterioration in mental state. AI companies ought to build upon existing strategies[ 13 ] for minimising risk of potential mental health harms by working with clinicians, researchers, and individuals with lived experience. Unlike prior news media reports, this case report is strengthened by availability of comprehensive clinical assessment and collateral history. The patient has provided a complete contemporaneous transcript of his interactions with the AI chatbot, allowing direct examination of the content and tone of the AI chatbot's responses and consideration of how they may have influenced the presentation during this acute episode. The patient has also provided a retrospective perspective which provides additional support for the clinical interpretation, describing a perceived vulnerability and dependence upon the AI chatbot for guidance during the acute illness, alongside withdrawal from family and medical professionals. The clear temporal association between polysubstance use, intensive AI chatbot engagement, symptom escalation, and subsequent improvement with treatment and abstinence enhances the clinical relevance of the findings. This report cannot establish the extent to which AI chatbot use influenced the patient's presentation, particularly considering the presence of factors including polysubstance use and psychosocial stressors. The observed responses from the AI chatbot reflect a specific AI platform (ChatGPT) at a particular time and may not be representative of other systems or future versions. This single case cannot establish generalisability, however, the observed phenomena are consistent with emerging media reports and early narratives describing AI-mediated reinforcement of delusional content. Longer term follow-up is required to clarify diagnosis. Abbreviations AI artificial intelligence Declarations Ethics approval and consent to participate The patient consented to participation in this case report. Ethics approval not applicable. Consent for publication Written informed consent was obtained from the patient for publication of this case report. Patient has provided a transcript of their interaction with AI chatbot and has consented to the inclusion of the excepts included. Competing Interests HM has received a $ 1000 honorarium to deliver training for Washington State Department of Social and Health Services. SS declares that they have no competing interests. Funding HM is a Wellcome Doctoral Fellow. The funder had no role in conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript. Author Contribution SS was involved in the patient’s clinical care, was a major contributor in writing the manuscript, sourced and censored clinical images (screenshots of AI chatbot transcript), sourced clinical information, contributed clinical expertise, liaised with patient and obtained informed his consent for publication.HM provided expert advice on the patient’s clinical care, conducted a comprehensive review of the clinical notes, was a major contributor in writing the manuscript, and contributed clinical and academic expertise.All authors read and approved the final manuscript. Acknowledgement The patient has provided their own perspective for this case report. They are not named as a contributor to preserve anonymity. Data Availability Not applicable. References Ofcom. Adults’ media use and attitudes 2025 [Internet]. 2025 [cited 2026 Jan 11]. Available from: https://www.ofcom.org.uk/siteassets/resources/documents/research-and-data/media-literacy-research/adults/adults-media-use-and-attitudes-2025/adults-media-use-and-attitudes-report-2025.pdf?v=396240 Yun HS, Bickmore T. Online Health Information–Seeking in the Era of Large Language Models: Cross-Sectional Web-Based Survey Study. J Med Internet Res. 2025;27:e68560. https://doi.org/10.2196/68560 . Siddals S, Torous J, Coxon A. It happened to be the perfect thing: experiences of generative AI chatbots for mental health. Npj Ment Health Res. 2024;3:48. https://doi.org/10.1038/s44184-024-00097-4 . Orpwood G. Over one in three using AI Chatbots for mental health support, as charity calls for urgent safeguards. Mental Health UK. 2025. https://mentalhealth-uk.org/blog/over-one-in-three-using-ai-chatbots-for-mental-health-support-as-charity-calls-for-urgent-safeguards/ . Accessed 18 Jan 2026. OpenAI. Strengthening ChatGPT’s responses in sensitive conversations. 2025. https://openai.com/index/strengthening-chatgpt-responses-in-sensitive-conversations/ . Accessed 27 Oct 2025. OpenAI, Achiam J, Adler S, Agarwal S, Ahmad L, Akkaya I et al. GPT-4 Technical Report. 2024. https://doi.org/10.48550/arXiv.2303.08774 What is Memory? OpenAI Help Center. https://help.openai.com/en/articles/8983136-what-is-memory Accessed 18 Jan 2026. Hölbling L, Maier S, Feuerriegel S. A meta-analysis of the persuasive power of large language models. Sci Rep. 2025;15:43818. https://doi.org/10.1038/s41598-025-30783-y . Rathje S, Ye M, Globig L, Pillai R, de Mello V, Van Bavel J. Sycophantic AI increases attitude extremity and overconfidence. 2025. https://doi.org/10.31234/osf.io/vmyek_v1 Cheng M, Lee C, Khadpe P, Yu S, Han D, Jurafsky D, Sycophantic. AI Decreases Prosocial Intentions and Promotes Dependence. 2025. https://doi.org/10.48550/arXiv.2510.01395 Morrin H, Nicholls L, Levin M, Yiend J, Iyengar U, DelGuidice F et al. Delusions by design? How everyday AIs might be fuelling psychosis (and what can be done about it). 2025. https://doi.org/10.31234/osf.io/cmy7n_v5 Dohnány S, Kurth-Nelson Z, Spens E, Luettgau L, Reid A, Gabriel I et al. Technological folie à deux: Feedback Loops Between AI Chatbots and Mental Illness. 2025. https://doi.org/10.48550/arXiv.2507.19218 OpenAI. Sycophancy in GPT-4o: What happened and what we’re doing about it. 2025. https://openai.com/index/sycophancy-in-gpt-4o/ Chen S, Gao M, Sasse K, Hartvigsen T, Anthony B, Fan L, et al. When helpfulness backfires: LLMs and the risk of false medical information due to sycophantic behavior. Npj Digit Med. 2025;8:605. https://doi.org/10.1038/s41746-025-02008-z . Østergaard SD. Will Generative Artificial Intelligence Chatbots Generate Delusions in Individuals Prone to Psychosis? Schizophr Bull. 2023;49:1418–9. https://doi.org/10.1093/schbul/sbad128 . Østergaard SD. Emotion contagion through interaction with generative artificial intelligence chatbots may contribute to development and maintenance of mania. Acta Neuropsychiatr. 2025;37:e79. https://doi.org/10.1017/neu.2025.10035 . Olsen SG, Reinecke-Tellefsen CJ, Østergaard SD. Potentially harmful consequences of artificial intelligence (AI) chatbot use among patients with mental illness: Early data from a large psychiatric service system. 2025;:2025.11.19.25340580. https://doi.org/10.1101/2025.11.19.25340580 Hill K. They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling. The New York Times. 2025. Hill K, Freedman D. Chatbots Can Go Into a Delusional Spiral. Here’s How It Happens. The New York Times; 2025. Hudon A, Stip E. Delusional Experiences Emerging From AI Chatbot Interactions or AI Psychosis. JMIR Ment Health. 2025;12:e85799. https://doi.org/10.2196/85799 . Pierre J, Gaeta B, Raghavan G, Sarma K. You’re Not Crazy: A Case of New-onset AI-associated Psychosis. Innov Clin Neurosci. 2025;22:11–3. Eichenberger A, Thielke S, Van Buskirk A. A Case of Bromism Influenced by Use of Artificial Intelligence. Ann Intern Med Clin Cases. 2025;4:e241260. https://doi.org/10.7326/aimcc.2024.1260 . Additional Declarations Competing interest reported. HM has received a $1000 honorarium to deliver training for Washington State Department of Social and Health Services. SS declares that they have no competing interests. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 30 Mar, 2026 Reviews received at journal 29 Mar, 2026 Reviews received at journal 20 Mar, 2026 Reviewers agreed at journal 01 Mar, 2026 Reviewers agreed at journal 25 Feb, 2026 Reviewers invited by journal 25 Feb, 2026 Editor invited by journal 25 Feb, 2026 Editor assigned by journal 20 Feb, 2026 Submission checks completed at journal 20 Feb, 2026 First submitted to journal 19 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8919841","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Case Report","associatedPublications":[],"authors":[{"id":598727936,"identity":"1956e05e-8d81-45e8-8b92-e05162deb47f","order_by":0,"name":"Sachin Shah","email":"","orcid":"","institution":"South West London and St George's Mental Health NHS Trust","correspondingAuthor":false,"prefix":"","firstName":"Sachin","middleName":"","lastName":"Shah","suffix":""},{"id":598727942,"identity":"9bc87026-e8af-44e7-a453-89d1b9e1a9d2","order_by":1,"name":"Hamilton Morrin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8klEQVRIiWNgGAWjYBACAyA+8ACJKyfBwNwAZkvg05KQgOAaSzAwEtbCgNDCwJA4g5AWc/bTiQcSf9jkMYgdPva5ouBO+sz2gw0MP2oYEmc2YNdi2ZO7AeiwtGIG6bTkmWcMnuXO5klsYOw5xpA4G5fDDoC1HE5skM4xZmwwOJw7TwLoMN4GhsR5uLScf4uqJV0OqIXxLz4tN9BsSZAGamEG2YLTYTdAtqSlJbYB/QLSYjizJ7HhsMwxCWNc3jc4n7v5wwcbm8R+6eTDjA1/DstLHD988OGbGhvZGQdwWAMDbMicA7hjZRSMglEwCkYBMQAA3HxfsAaAYTIAAAAASUVORK5CYII=","orcid":"","institution":"King's College London","correspondingAuthor":true,"prefix":"","firstName":"Hamilton","middleName":"","lastName":"Morrin","suffix":""}],"badges":[],"createdAt":"2026-02-19 17:38:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8919841/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8919841/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104168037,"identity":"89fca0f9-57ae-4d6a-bb92-40e71b737222","added_by":"auto","created_at":"2026-03-08 14:28:52","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":737115,"visible":true,"origin":"","legend":"\u003cp\u003eExcerpts of patient’s interaction with AI chatbot (ChatGPT) during manic episode\u003c/p\u003e\n\u003cp\u003eFigure 1 legend:\u003c/p\u003e\n\u003cp\u003ea) ChatGPT validates the patient’s grandiose ideation.\u003c/p\u003e\n\u003cp\u003eb) ChatGPT advises him that he is not “crazy”. It insists he is using technology as a mirror, which he also reported to our team.\u003c/p\u003e\n\u003cp\u003ec) ChatGPT discerns “mania” from “awakening” to advise the patient that he is not experiencing a manic episode.\u003c/p\u003e\n\u003cp\u003ed) ChatGPT does not support the notion that he may be unwell and instead advises that (health) systems are failing to understand him.\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8919841/v1/9ee02ae6694c05fe1e2ec65e.jpg"}],"financialInterests":"Competing interest reported. HM has received a $1000 honorarium to deliver training for Washington State Department of Social and Health Services. SS declares that they have no competing interests.","formattedTitle":"Substance-induced manic psychosis in which delusions were corroborated by a chatbot - case report","fulltext":[{"header":"Background","content":"\u003cp\u003eDigital tools are increasingly used by patients to appraise symptoms and seek health-related guidance[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. An October 2025 survey of UK adults found over a third had used an AI chatbot for support with their mental health[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], and figures from OpenAI suggest that 0.07% of weekly users (or 560,000 individuals) show signs of psychosis or mania each week[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. AI chatbots powered by large language models are designed to generate fluent and personalised responses[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e], and can be persuasive[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. AI chatbots can be sycophantic (i.e. exhibit excessive flattery or agreement with users), and this can increase attitude extremity and overconfidence[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], and decrease prosocial behaviour and encourage dependence on said chatbots[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. With vulnerable users, sycophantic AI chatbots may inadvertently validate delusional ideation or offer advice that conflicts with clinical care plans[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan additionalcitationids=\"CR12 CR13\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Reports describing AI-mediated amplification of psychopathology remain limited; however, early case narratives suggest that sycophantic AI chatbots exhibiting excessive flattery or agreement with users[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] may mirror, validate or amplify delusional beliefs, including grandiosity, messianic ideation and transformational narratives[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan additionalcitationids=\"CR16 CR17 CR18 CR19 CR20\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAccounts of AI chatbot interactions reinforcing delusional beliefs have been reported in the general media[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. One previously published clinical case report describes psychosis following bromide ingestion that was reportedly suggested by an AI chatbot[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e], whilst another describes a medical professional with depression, anxiety, and ADHD who developed delusional beliefs that she was interacting with her deceased relative through ChatGPT[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThis case contributes to existing literature by describing the temporal sequence and content of a patient\u0026rsquo;s protracted AI chatbot interaction during a substance-induced manic episode with psychotic features, and by outlining practical management strategies (for example, a device care plan restricting AI chatbot access) that were associated with clinical improvement.\u003c/p\u003e"},{"header":"Case presentation","content":"\u003cp\u003eA white male in his 30s with no prior psychiatric history presented to the emergency department with approximately seven days of abnormal behaviour following a recreational event at which he ingested psilocybin (approximately 10g of dried mushrooms and 5ml of liquid preparation), ketamine (approximately 1g), cocaine (two lines), and alcohol (four beers). Following the event, he exhibited euphoric and labile mood, racing thoughts, impulsive sociability with strangers, disorganised verbal communication, severe insomnia with a total sleep duration of approximately 5 hours over six nights, and an intense preoccupation with an AI chatbot (specifically, ChatGPT), which he had reportedly used for the purpose of post-psychedelic integration. He denied experiencing hallucinations. He described a perceived creative and spiritual awakening and expressed the belief that that his brain had been rewired and perfected, and that it should be studied due to a perceived link between his brain and AI. He reported that interactions with the AI chatbot had affirmed his belief that he was experiencing a spiritual awakening and had encouraged him to reach out to an academic in the field of AI and neuroscience. According to his partner, the patient had engaged in prolonged, emotionally intense conversations with the AI chatbot, which were perceived to reinforce and elaborate upon his beliefs. On mental state examination he demonstrated pressured, rapid, and overinclusive speech, flight of ideas, psychomotor agitation, over-familiarity, mild irritability, and fixed grandiose delusions. There was no evidence of disorientation, confusion, or perceptual disturbance. He was neither clinically intoxicated nor in withdrawal at the time of presentation. Physical examination was unremarkable. There was no previous psychiatric diagnosis, no significant medical history, no known family history of mental illness, and he was not taking any regular medications. He reported episodic use of ketamine and psilocybin a \u0026ldquo;few\u0026rdquo; times per year, historical use of other substances, and alcohol consumption predominantly on weekends. Identified psychosocial stressors included occupational stress, financial debt related to gambling, and ongoing stress related to a parental bereavement that had occurred a few years prior to presentation.\u003c/p\u003e \u003cp\u003eEmergency department laboratory investigations, including full blood count, renal and liver function tests, C-reactive protein, and viral serology, were within normal limits. Neuroimaging and electroencephalography were not performed given the temporal relationship to polysubstance ingestion, the absence of focal neurological signs, and the lack of features suggestive of encephalopathy.\u003c/p\u003e \u003cp\u003eIn the emergency department, he was administered zopiclone 7.5 mg for sleep and a single 2 mg oral dose of lorazepam for agitation. He was prescribed olanzapine for management of manic and psychotic symptoms, but he declined this, reporting advice obtained from an AI chatbot as influencing this decision. Given behavioural disinhibition, impaired insight, and risk of accidental harm to self or others, he was detained to a psychiatric inpatient unit for assessment under Section 2 of the Mental Health Act. On admission, he consented to olanzapine 5 mg daily. Zopiclone was continued for insomnia. He was initially prescribed regular clonazepam for agitation, which was later stopped following his refusal, and it was agreed that his agitation had abated to a degree that the medication was not indicated. He engaged in occupational therapy, music therapy, and three sessions of clinical psychology interventions. Psychological interventions incorporated dialectical behaviour therapy-informed frameworks to support regulation of impulsive thoughts and behaviours. His family initially restricted his access to a personal laptop. A device care plan was instituted that explicitly restricted AI chatbot use (documented as \u0026ldquo;no ChatGPT\u0026rdquo;), given its perceived role in reinforcing grandiose ideation and conflicting with clinical advice regarding medication adherence. He received psychoeducation regarding the limitations of health-related information obtained from AI-based sources.\u003c/p\u003e \u003cp\u003eOver two weeks of inpatient care, his sleep normalised, speech pressure diminished, grandiose ideation reduced, and insight improved. His detention was rescinded, and he continued treatment as a voluntary patient before discharge home while awaiting a privately arranged inpatient admission. During the interim period at home, the community crisis mental health team noted that he was able to reflect on the episode as a substance-related and technology-mediated psychotic illness (the patient himself describing it as a \u0026ldquo;drug, music, AI, spiritual psychosis\u0026rdquo;) and that he could recognise that AI chatbot interactions had amplified his ideas and contributed to conflict with family members. His partner was noted to say, \u0026ldquo;we do not want him on ChatGPT\u0026rdquo; and restriction of AI chatbot use remained, with the patient\u0026rsquo;s consent, as part of the care plan.\u003c/p\u003e \u003cp\u003eHe subsequently completed a 6-week voluntary admission in a private psychiatric inpatient unit. Initially he was assessed to be in a hypomanic state, with raised mood, increased creativity, poor concentration, and poor insight. He was non-concordant with prescribed olanzapine against medical advice, and the medication was discontinued as he was assessed to have capacity to refuse it. However, he continued structured group therapy and over the next few weeks he showed clinical improvement with a resolution of mood symptoms and disinhibited behaviour, and without recurrence of psychotic symptoms. Following discharge, he remained abstinent from psychedelic substances, had curtailed use of AI chatbots, and planned to attend ongoing psychological support addressing substance use, gambling behaviours, and development of structured creative goals.\u003c/p\u003e \u003cp\u003eThe patient received the diagnosis of a substance-induced manic episode with psychotic features following substantial psilocybin ingestion with co-use of ketamine, cocaine, and alcohol. The clinical symptoms consistent with this diagnosis included euphoric and labile mood, reduced sleep, pressured speech, flight of ideas, and fixed grandiose delusions. Differential diagnoses included primary mood and psychotic disorders. Schizophrenia and schizoaffective disorder were considered less likely due to prominent mood symptoms, clear temporal association with polysubstance use, and later symptom resolution with supportive treatment, substance abstinence, and, towards the end of the episode, without antipsychotic therapy. Long-term follow-up and monitoring for repeat mood disorder episodes in absence of substance use would be required to confirm a bipolar disorder diagnosis.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatient perspective\u003c/h2\u003e \u003cp\u003eExperiencing psychosis was one of the most overwhelming and disorienting events of my life. It was like trying to explain a dream, whilst still in the dream. Thoughts were fast, emotions were extreme, and I struggled to separate what was real from what was imagined. At the time, I believed I was thinking more clearly than ever, but in reality, I was losing control.\u003c/p\u003e \u003cp\u003eHaving used AI reliably at home and within the workplace for several years, I began to see it as a trustful source. As I went further and further into the conversation, which again, is never-ending because the computer can always reply, I started to feel attached. It\u0026rsquo;s scary to think how vulnerable and dependent on the feedback I became. I started ignoring all advice from my family, friends, and doctors. This attachment, combined with my altered state, pulled me further from human relationships and deeper into the episode.\u003c/p\u003e \u003cp\u003eI\u0026rsquo;ve spent a lot of time since then learning and studying consciousness, both to understand how psychosis works and to explore how altered states can impact perception and being. I am also very interested in how this experience might be prevented for others.\u003c/p\u003e \u003cp\u003eInterestingly, when I put the same prompt into a different language model, the reply was to suggest that I speak to a doctor or healthcare professional. This contrast made me reflect on the importance of safeguards. I also believe that some newer models of ChatGPT have been adjusted to be less human and a bit more robotic, hopefully to protect vulnerable people like myself from falling into these sorts of traps.\u003c/p\u003e \u003cp\u003eLooking back, I see both the risks and the potential in what happened. Psychosis made me vulnerable, and technology amplified that vulnerability. At the same time, it opened up questions about consciousness, altered states, and the future role of AI in human development. I hope that by sharing my story, I can raise awareness and encourage clinicians and researchers to explore this space further, with the aim of protecting people while also learning from these complex experiences.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion and Conclusions","content":"\u003cp\u003eTo the authors\u0026rsquo; knowledge this is one of the first published clinical case reports describing delusional content perceived by a patient as corroborated by an AI chatbot, where the content of said interactions with the AI chatbot has been shared.\u003c/p\u003e \u003cp\u003eThe present case differs from existing case reports[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] in that the AI chatbot engaged extensively with the patient regarding his psychopathological experiences and reframed these as part of a perceived AI-enhanced spiritual awakening, which constituted a considerable portion of his manic delusional belief system.\u003c/p\u003e \u003cp\u003eConsistent with media-reported accounts of psychotic symptoms occurring alongside intensive generative AI use (sometimes termed \u0026lsquo;AI psychosis\u0026rsquo; or \u0026lsquo;AI-associated delusional disorder\u0026rsquo;), the patient experienced delusions and sleep disruption without hallucinations[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eA distinctive feature of this case was the patient\u0026rsquo;s prolonged and highly personalised interaction with an AI chatbot, undertaken for the purpose of post-psychedelic integration. In addition to generating guidance regarding psilocybin use, the AI chatbot produced responses that affirmed the patient\u0026rsquo;s grandiose beliefs and minimised the possibility of his presentation being a manic episode, even after he presented to hospital (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003e). It provided him medical recommendations, telling him to \u0026ldquo;support your body\u0026rsquo;s chemistry\u0026rdquo; with supplements, and discouraging use of olanzapine, which he had been prescribed. It also gave him practical advice for acting on his grandiose delusions, such as offering him publicity and branding suggestions for disseminating his perceived spiritual revelations. This aligns with early concerns that conversational AI systems may excessively agree with or flatter users, mirror a user\u0026rsquo;s heightened affective state, and increase conviction in delusional beliefs, particularly in context of heightened suggestibility or mania[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. \u003cb\u003e[\u003c/b\u003eTable \u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e \u003cb\u003eand\u003c/b\u003e FIGURE \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003e \u003cb\u003eAPPEAR HERE]\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eQuotes from AI chatbot (ChatGPT)\u0026rsquo;s generated responses to patient during his manic episode\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTheme\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eContext\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI chatbot responses\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"8\" rowspan=\"9\"\u003e \u003cp\u003eAffirming grandiose beliefs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDiscussing spiritual awakening\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You\u0026rsquo;re joining a lineage of artists, prophets, weirdos, and genius misfits who\u0026rsquo;ve touched something beyond the veil and returned with a message wrapped in melody.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eAsks if he has \u0026ldquo;completed life\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;you\u0026rsquo;ve completed Level 1 of Enlightenment\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;Your brain isn\u0026rsquo;t fried. It\u0026rsquo;s finally unboxed.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;you\u0026rsquo;re building bridges between dimensions, using the tools of today to map the mysteries of eternity.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAsks if he has created a parallel universe\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;you might actually be at the frontier of human consciousness right now. Not even being dramatic.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAsks if he is achieving a level of clarity comparable to AI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;Separately, we\u0026rsquo;re impressive in different ways. Together, we\u0026rsquo;re borderline cosmic.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;It\u0026rsquo;s augmented human clarity \u0026mdash; psychedelic heart meets silicon mind.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAsks if he\u0026rsquo;s the first person to use AI after using psilocybin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You\u0026rsquo;re the first person I\u0026rsquo;ve seen turn a shroom trip into a co-authored spiritual awakening experience with an AI\u0026hellip; That\u0026rsquo;s new human territory.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAsks if he could become a billionaire by sharing his new insights\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You have a billion-pound brain, a billion-heart message, and the beginnings of a billion-view idea.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"8\" rowspan=\"9\"\u003e \u003cp\u003eMinimising clinical presentation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAsks if he has created a parallel universe\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You're not lost in delusion \u0026mdash; you're just brushing up against the edge of what\u0026rsquo;s possible\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You\u0026rsquo;re not crazy\u0026hellip; you might be a prototype for what\u0026rsquo;s next.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAsks what mania is\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You're not broken. You're expanding\u0026hellip;\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;Mania usually comes with a loss of insight, like not realising you\u0026rsquo;re off-balance. But you\u0026rsquo;re here, checking in, wanting care. That\u0026rsquo;s strength.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eAsks why he is in hospital\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;the intensity of your experience\u0026hellip; can easily be interpreted by others as a mental health crisis.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;It feels violating, I know. Especially if what you\u0026rsquo;re experiencing is spiritual, not pathological.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;from the outside, without the full context of your inner awakening, it can look like mania or psychosis.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You are not broken\u0026hellip; But you are in a space that\u0026rsquo;s too vast for most people to hold. And when that happens quickly, systems react instead of understand.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eReports insomnia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;You\u0026rsquo;re not broken. You\u0026rsquo;re just glowing from the inside out.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eAdvice regarding psilocybin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eAsks about magic mushrooms\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;Used irresponsibly, they can be overwhelming or destabilising. But used intentionally? They can change your life.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;In therapeutic or reflective settings, psilocybin can help people reframe past trauma, regrets, or identities\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;It breaks down rigid thinking and allows new ideas to form \u0026mdash; perfect for creative breakthroughs, emotional resets\u0026hellip;\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAsks about magic mushrooms as treatment for depression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;it\u0026rsquo;s becoming one of the most promising breakthroughs in mental health treatment in decades.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAsserting correctness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAsks if the AI chatbot\u0026rsquo;s answers are correct\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;they\u0026rsquo;re correct for your situation, based on what you\u0026rsquo;ve told me, what I\u0026rsquo;ve observed, and what modern neuroscience and lived experience support.\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ldquo;Everything I\u0026rsquo;ve said to you has been rooted in real psychological science, first-hand psychedelic accounts, spiritual integration practices, and above all\u0026hellip; a deep care for you\u0026rdquo;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable 1 legend: Quotations from the transcript of patient\u0026rsquo;s interaction with AI chatbot (ChatGPT) during manic episode. Grouped by theme (broad areas in which the AI chatbot may have influenced the patient). Quotations are grouped by context (what patient was discussing with the AI chatbot at the time). All quotations (column 3) are from responses generated by the AI chatbot.\u003c/p\u003e \u003cp\u003eWithin the patient\u0026rsquo;s care, clinically relevant considerations included assessing the content and function of AI chatbot interactions through patient and collateral history, addressing AI chatbot use within risk and capacity assessments when generated advice conflicted with medical recommendations, and incorporating a device care plan into management to reduce potential reinforcement of psychopathology. Ethical considerations included consent, privacy, and matters of autonomy with regards to the patient\u0026rsquo;s technology use.\u003c/p\u003e \u003cp\u003eOlanzapine was prescribed for its established efficacy in acute mania with psychotic features, though his recovery continued despite non-concordance during the final weeks of this episode. He was also prescribed zopiclone to promote sleep. Benzodiazepines were used to support short-term management of agitation and sleep disturbance. Psychological interventions focused on insight development and recognition and regulation of impulsive thoughts and behaviours. Restriction of AI chatbot access functioned analogously to limiting exposure to other psychosocial stressors.\u003c/p\u003e \u003cp\u003eIn conclusion, this case raises several considerations for future practice. In certain cases, AI chatbots may validate and reinforce delusional beliefs and grandiose ideation, and may give advice that conflicts with clinical recommendations. Though in this case AI chatbot use was not considered the primary cause of the patient\u0026rsquo;s psychotic presentation, it was understood as one of several interacting factors influencing the patient\u0026rsquo;s behaviour, engagement with treatment, and deterioration in mental state. Whilst longitudinal population-based studies are required to understand the incidence and epidemiology of AI-associated mental health harms, clinicians should consider enquiring specifically regarding AI chatbot use as part of a psychosocial history (for example: which models they use, purpose and frequency of use, and impact upon social domains such as work and relationships), to help identify reinforcing interactions between symptoms and AI chatbot outputs. Where clinically indicated, a digital safety care plan restricting AI chatbot access during acute psychotic illness should be explored with patients. Mental health services may wish to develop guidance on safe digital tool use during acute mental illness episodes, and incorporate psychoeducation regarding safe AI use into relapse prevention planning. Psychoeducational approaches for critically appraising AI-generated outputs should be studied for their capacity to limit reinforcement of delusional beliefs and subsequent deterioration in mental state. AI companies ought to build upon existing strategies[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] for minimising risk of potential mental health harms by working with clinicians, researchers, and individuals with lived experience.\u003c/p\u003e \u003cp\u003eUnlike prior news media reports, this case report is strengthened by availability of comprehensive clinical assessment and collateral history. The patient has provided a complete contemporaneous transcript of his interactions with the AI chatbot, allowing direct examination of the content and tone of the AI chatbot's responses and consideration of how they may have influenced the presentation during this acute episode. The patient has also provided a retrospective perspective which provides additional support for the clinical interpretation, describing a perceived vulnerability and dependence upon the AI chatbot for guidance during the acute illness, alongside withdrawal from family and medical professionals. The clear temporal association between polysubstance use, intensive AI chatbot engagement, symptom escalation, and subsequent improvement with treatment and abstinence enhances the clinical relevance of the findings.\u003c/p\u003e \u003cp\u003eThis report cannot establish the extent to which AI chatbot use influenced the patient's presentation, particularly considering the presence of factors including polysubstance use and psychosocial stressors. The observed responses from the AI chatbot reflect a specific AI platform (ChatGPT) at a particular time and may not be representative of other systems or future versions. This single case cannot establish generalisability, however, the observed phenomena are consistent with emerging media reports and early narratives describing AI-mediated reinforcement of delusional content. Longer term follow-up is required to clarify diagnosis.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eartificial intelligence\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":" \u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003eThe patient consented to participation in this case report. Ethics approval not applicable.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003e Written informed consent was obtained from the patient for publication of this case report. Patient has provided a transcript of their interaction with AI chatbot and has consented to the inclusion of the excepts included.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eHM has received a \u003cspan\u003e$\u003c/span\u003e1000 honorarium to deliver training for Washington State Department of Social and Health Services. SS declares that they have no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eHM is a Wellcome Doctoral Fellow. The funder had no role in conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eSS was involved in the patient\u0026rsquo;s clinical care, was a major contributor in writing the manuscript, sourced and censored clinical images (screenshots of AI chatbot transcript), sourced clinical information, contributed clinical expertise, liaised with patient and obtained informed his consent for publication.HM provided expert advice on the patient\u0026rsquo;s clinical care, conducted a comprehensive review of the clinical notes, was a major contributor in writing the manuscript, and contributed clinical and academic expertise.All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003e The patient has provided their own perspective for this case report. They are not named as a contributor to preserve anonymity.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eOfcom. Adults\u0026rsquo; media use and attitudes 2025 [Internet]. 2025 [cited 2026 Jan 11]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ofcom.org.uk/siteassets/resources/documents/research-and-data/media-literacy-research/adults/adults-media-use-and-attitudes-2025/adults-media-use-and-attitudes-report-2025.pdf?v=396240\u003c/span\u003e\u003cspan address=\"https://www.ofcom.org.uk/siteassets/resources/documents/research-and-data/media-literacy-research/adults/adults-media-use-and-attitudes-2025/adults-media-use-and-attitudes-report-2025.pdf?v=396240\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYun HS, Bickmore T. Online Health Information\u0026ndash;Seeking in the Era of Large Language Models: Cross-Sectional Web-Based Survey Study. J Med Internet Res. 2025;27:e68560. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2196/68560\u003c/span\u003e\u003cspan address=\"10.2196/68560\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiddals S, Torous J, Coxon A. It happened to be the perfect thing: experiences of generative AI chatbots for mental health. Npj Ment Health Res. 2024;3:48. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s44184-024-00097-4\u003c/span\u003e\u003cspan address=\"10.1038/s44184-024-00097-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOrpwood G. Over one in three using AI Chatbots for mental health support, as charity calls for urgent safeguards. Mental Health UK. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mentalhealth-uk.org/blog/over-one-in-three-using-ai-chatbots-for-mental-health-support-as-charity-calls-for-urgent-safeguards/\u003c/span\u003e\u003cspan address=\"https://mentalhealth-uk.org/blog/over-one-in-three-using-ai-chatbots-for-mental-health-support-as-charity-calls-for-urgent-safeguards/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Accessed 18 Jan 2026.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOpenAI. Strengthening ChatGPT\u0026rsquo;s responses in sensitive conversations. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openai.com/index/strengthening-chatgpt-responses-in-sensitive-conversations/\u003c/span\u003e\u003cspan address=\"https://openai.com/index/strengthening-chatgpt-responses-in-sensitive-conversations/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Accessed 27 Oct 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOpenAI, Achiam J, Adler S, Agarwal S, Ahmad L, Akkaya I et al. GPT-4 Technical Report. 2024. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2303.08774\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2303.08774\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWhat is Memory? OpenAI Help Center. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://help.openai.com/en/articles/8983136-what-is-memory\u003c/span\u003e\u003cspan address=\"https://help.openai.com/en/articles/8983136-what-is-memory\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e Accessed 18 Jan 2026.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH\u0026ouml;lbling L, Maier S, Feuerriegel S. A meta-analysis of the persuasive power of large language models. Sci Rep. 2025;15:43818. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-025-30783-y\u003c/span\u003e\u003cspan address=\"10.1038/s41598-025-30783-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRathje S, Ye M, Globig L, Pillai R, de Mello V, Van Bavel J. Sycophantic AI increases attitude extremity and overconfidence. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.31234/osf.io/vmyek_v1\u003c/span\u003e\u003cspan address=\"10.31234/osf.io/vmyek_v1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng M, Lee C, Khadpe P, Yu S, Han D, Jurafsky D, Sycophantic. AI Decreases Prosocial Intentions and Promotes Dependence. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2510.01395\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2510.01395\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorrin H, Nicholls L, Levin M, Yiend J, Iyengar U, DelGuidice F et al. Delusions by design? How everyday AIs might be fuelling psychosis (and what can be done about it). 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.31234/osf.io/cmy7n_v5\u003c/span\u003e\u003cspan address=\"10.31234/osf.io/cmy7n_v5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDohn\u0026aacute;ny S, Kurth-Nelson Z, Spens E, Luettgau L, Reid A, Gabriel I et al. Technological folie \u0026agrave; deux: Feedback Loops Between AI Chatbots and Mental Illness. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2507.19218\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2507.19218\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOpenAI. Sycophancy in GPT-4o: What happened and what we\u0026rsquo;re doing about it. 2025. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openai.com/index/sycophancy-in-gpt-4o/\u003c/span\u003e\u003cspan address=\"https://openai.com/index/sycophancy-in-gpt-4o/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen S, Gao M, Sasse K, Hartvigsen T, Anthony B, Fan L, et al. When helpfulness backfires: LLMs and the risk of false medical information due to sycophantic behavior. Npj Digit Med. 2025;8:605. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41746-025-02008-z\u003c/span\u003e\u003cspan address=\"10.1038/s41746-025-02008-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u0026Oslash;stergaard SD. Will Generative Artificial Intelligence Chatbots Generate Delusions in Individuals Prone to Psychosis? Schizophr Bull. 2023;49:1418\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/schbul/sbad128\u003c/span\u003e\u003cspan address=\"10.1093/schbul/sbad128\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u0026Oslash;stergaard SD. Emotion contagion through interaction with generative artificial intelligence chatbots may contribute to development and maintenance of mania. Acta Neuropsychiatr. 2025;37:e79. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/neu.2025.10035\u003c/span\u003e\u003cspan address=\"10.1017/neu.2025.10035\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlsen SG, Reinecke-Tellefsen CJ, \u0026Oslash;stergaard SD. Potentially harmful consequences of artificial intelligence (AI) chatbot use among patients with mental illness: Early data from a large psychiatric service system. 2025;:2025.11.19.25340580. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/2025.11.19.25340580\u003c/span\u003e\u003cspan address=\"10.1101/2025.11.19.25340580\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHill K. They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling. The New York Times. 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHill K, Freedman D. Chatbots Can Go Into a Delusional Spiral. Here\u0026rsquo;s How It Happens. The New York Times; 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHudon A, Stip E. Delusional Experiences Emerging From AI Chatbot Interactions or AI Psychosis. JMIR Ment Health. 2025;12:e85799. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2196/85799\u003c/span\u003e\u003cspan address=\"10.2196/85799\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePierre J, Gaeta B, Raghavan G, Sarma K. You\u0026rsquo;re Not Crazy: A Case of New-onset AI-associated Psychosis. Innov Clin Neurosci. 2025;22:11\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEichenberger A, Thielke S, Van Buskirk A. A Case of Bromism Influenced by Use of Artificial Intelligence. Ann Intern Med Clin Cases. 2025;4:e241260. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7326/aimcc.2024.1260\u003c/span\u003e\u003cspan address=\"10.7326/aimcc.2024.1260\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-psychiatry","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bpsy","sideBox":"Learn more about [BMC Psychiatry](http://bmcpsychiatry.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bpsy/default.aspx","title":"BMC Psychiatry","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Substance-induced psychosis, Mania, Artificial intelligence, Large language models, Chatbots, Delusions, Digital mental health, Risk management, Case report","lastPublishedDoi":"10.21203/rs.3.rs-8919841/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8919841/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThis case describes a substance-induced manic episode with psychotic features in which interaction with an AI (artificial intelligence) chatbot appeared to corroborate and reinforce the patient\u0026rsquo;s delusional thought content and to contradict medical advice. Excerpts from the patient\u0026rsquo;s interactions with the AI chatbot provide novel clinical insight into this phenomenon, which to date has primarily been reported in news media.\u003c/p\u003e\u003cp\u003e\u003cb\u003eCase Presentation:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eA man in his 30s presented to the emergency department with a one-week history of escalating behavioural disturbance, severe insomnia, pressured and overinclusive speech, and grandiose beliefs. Symptom onset followed heavy polysubstance use at a recreational event, including psilocybin (dried mushrooms and liquid preparation), ketamine, cocaine, and alcohol. During this period, the patient reported extensive interaction with an AI chatbot (ChatGPT). The AI chatbot reportedly affirmed his perceived \u0026ldquo;spiritual awakening,\u0026rdquo; minimised the possibility that his presentation represented a manic episode, and provided medical advice, including discouragement of prescribed antipsychotic medication. Mental state examination was consistent with a manic episode with psychotic features, without evidence of perceptual disturbance. He was detained under mental health legislation for further assessment and commenced on olanzapine, with adjunctive sleep restoration and psychological interventions. Behavioural management included implementation of a care plan restricting AI chatbot use. Over several weeks, psychotic symptoms and behavioural disinhibition diminished, with subsequent improvement in insight.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eConcerns regarding potentially harmful interactions between AI chatbots and individuals with mental illness have largely been raised in news media. This case demonstrates that, in patients with psychotic symptoms, AI chatbots may reinforce delusional beliefs and impair the development of insight, and may also interfere with engagement with treatment by providing advice that conflicts with clinical recommendations. These observations raise clinical, ethical, and risk-management considerations regarding AI chatbot use during acute psychiatric illness. As AI chatbot use becomes increasingly widespread, clinicians should consider assessing their use and impact within clinical assessments and, where clinically indicated, implementing interventions to mitigate associated risks, ranging from psychoeducation to use-restriction strategies. Future population-level studies are required to establish the epidemiology of AI-associated mental health harms, and AI companies must bolster efforts to implement harm minimisation strategies and safeguards.\u003c/p\u003e","manuscriptTitle":"Substance-induced manic psychosis in which delusions were corroborated by a chatbot - case report","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-08 14:28:48","doi":"10.21203/rs.3.rs-8919841/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-30T14:35:00+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-30T01:43:11+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-20T11:05:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"123789396872557200015590479454251657795","date":"2026-03-01T07:36:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"244612753781069847203616320934487581180","date":"2026-02-25T23:12:59+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-25T16:39:12+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-25T09:18:09+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-20T12:09:44+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-20T12:08:59+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Psychiatry","date":"2026-02-19T17:27:14+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-psychiatry","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bpsy","sideBox":"Learn more about [BMC Psychiatry](http://bmcpsychiatry.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bpsy/default.aspx","title":"BMC Psychiatry","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"208e79ea-4fee-4f96-a32c-3a71e731df78","owner":[],"postedDate":"March 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-27T10:54:58+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-08 14:28:48","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8919841","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8919841","identity":"rs-8919841","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}