ChatGPT and mental healthcare: balancing benefits with risks of harms
Open accessThe DOI matches its Crossref record; 12 of the 12 references it cites were found in a registry or an index.
Abstract
Against the global need for increased access to mental services, health organisations are looking to technological advances to improve the delivery of care and lower costs. Since November 2022, with the public launch of OpenAI's ChatGPT, the field of generative artificial intelligence (AI) has received expanding attention. Although generative AI itself is not new, technical advances and the increased accessibility of large language models (LLMs) (eg, OpenAI's GPT-4 and Google's Bard) suggest use of these tools could be clinically significant. LLMs are an application of generative AI technology that can summarise and generate content based on training on vast data sets. Unlike search engines, which provide internet links in response to typed entries, chatbots that rely on generative language models can simulate dialogue that resembles human conversations. We examine the potential promise and the risks of using LLMs in mental healthcare today, focusing on their scope to impact mental healthcare, including global equity in the delivery of care. Although we caution that LLMs should not be used to disintermediate mental health clinicians, we signal how-if carefully implemented-in the long term these tools could reap benefits for patients and health professionals.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Hallucinations in the (mental) health context may be caused by a lack of training data or outdated medical sources47 and lead to low-quality or wrong mental health advice.”
From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [47]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“These risks are not hypothetical: real‐world examples, such as Replika's backlash for generating inappropriate responses 194 , illustrate the consequences of insufficient safeguards in generative AI systems 195 .”
From Charting the evolution of artificial intelligence mental health chatbots from rule‐based systems to large language models: a systematic review (2025) · cites this paper as [195]
The cited source says what the claim attributes to it.
In the citing paper
“Blease and Torous describe that such a risk may emerge from LLM-CBs not exclusively trained on medical information, leading to inappropriate or inconsistent responses in a medical context47.”
From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [47]
The cited source says what the claim attributes to it.
In the citing paper
“Data security concerns (mentioned by 13 articles) stem from security risks relating to sensitive health data provided by users or by clinicians36–39,44,47,50,51,55.”
From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [47]
The cited source says what the claim attributes to it.
In the citing paper
“The search string was validated using three pre-selected target articles47,84,110.”
From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [47]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Ayers, J. W., Poliak, A., Dredze, M., Leas, E. C., Zhu, Z., Kelley, J. B., Faix, D. J., Goodman, A. M., Longhurst, C. A., Hogarth, M., et al. (2023) · JAMA Internal Medicine · doi:10.1001/jamainternmed.2023.1838 · Cited by 2,613
Artificial hallucinations in ChatGPT: implications in scientific writing
Alkaissi, H., & McFarlane, S. I. (2023) · Cureus · doi:10.7759/cureus.35179 · Cited by 1,540
The World Mental Health Report: Transforming mental health for all
Freeman, M. (2022) · World Psychiatry · doi:10.1002/wps.21018 · Cited by 1,472
Sharma, A., Lin, I. W., Miner, A. S., Atkins, D. C., & Althoff, T. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-022-00593-2 · Cited by 500
Accuracy of a Generative Artificial Intelligence Model in a Complex Diagnostic Challenge
Kanjee, Z., Crowe, B., & Rodman, A. (2023) · JAMA · doi:10.1001/jama.2023.8288 · Cited by 449 · cited 1×
“Encouragingly, preliminary studies show the promise of GPT-4 in generating accurate lists of differential diagnoses, including in complicated clinical cases, 7 suggesting their potential to facilitate brainstorming in diagnostic and treatment decision-making.”
BackgroundAI Chatbots, Health Privacy, and Challenges to HIPAA Compliance
Marks, M., & Haupt, C. E. (2023) · JAMA · doi:10.1001/jama.2023.9458 · Cited by 171 · cited 1×
“17 In the USA, there are already efforts to integrate generative AI services into electronic healthcare systems that comply with the privacy standards of the 1996 Health Insurance Portability and Account-ability Act.”
BackgroundThe Challenges for Regulating Medical Use of ChatGPT and Other Large Language Models
Minssen, T., Vayena, E., & Cohen, I. G. (2023) · JAMA · doi:10.1001/jama.2023.9651 · Cited by 168
What ChatGPT Tells Us about Gender: A Cautionary Tale about Performativity and Gender Biases in AI
Gross, N. (2023) · Social Sciences · doi:10.3390/socsci12080435 · Cited by 162
Should ChatGPT Be Biased? Challenges and Risks of Bias in Large Language Models
Ferrara, E. (2023) · doi:10.2139/ssrn.4627814 · Cited by 133
Blease, C., Locher, C., Leon-Carlyle, M., & Doraiswamy, M. (2020) · DIGITAL HEALTH · doi:10.1177/2055207620968355 · Cited by 121
Self-citation
Research works printed but not linked: 1
- AMA. ChatGPT and generative AI: what physicians should consider. American Medical Association; 2023. Available: https://www.ama-assn.org/system/files/chatgpt-what-physicians-should-consider.pdf [Accessed 11 Sep 2023].
Cited by
Generative artificial intelligence in primary care: an online survey of UK general practitioners
Blease, C. R., Locher, C., Gaab, J., Hägglund, M., & Mandl, K. D. (2024) · BMJ Health & Care Informatics · doi:10.1136/bmjhci-2024-101102 · Cited by 163
Self-citationHua, Y., Siddals, S., Ma, Z., Galatzer‐Levy, I., Xia, W., Hau, C., Na, H., Flathers, M., Linardon, J., Ayubcha, C., et al. (2025) · World Psychiatry · doi:10.1002/wps.21352 · Cited by 117
“These risks are not hypothetical: real-world examples, such as Replika ’s back-lash for generating inappropriate responses 194 , illustrate the consequences of insufficient safeguards in generative AI systems 195 .”
Self-citationUse of AI in Mental Health Care: Community and Mental Health Professionals Survey
Cross, S., Bell, I., Nicholas, J., Valentine, L., Mangelsdorf, S., Baker, S., Titov, N., & Alvarez-Jimenez, M. (2024) · JMIR Mental Health · doi:10.2196/60589 · Cited by 113
“These findings align with the broader discourse in general health care delivery on the integration of AI into care, where efficiency and productivity gains must be balanced with accuracy, reliability, and ethical considerations [20,31].”
BackgroundMethodInfluentialApplications of large language models in psychiatry: a systematic review
Omar, M., Soffer, S., Charney, A. W., Landi, I., Nadkarni, G. N., & Klang, E. (2024) · Frontiers in Psychiatry · doi:10.3389/fpsyt.2024.1422807 · Cited by 84
“Despite these capabilities, currently, LLMs do not replace human therapists (8, 18, 19).”
BackgroundCognitive offloading or cognitive overload? How AI alters the mental architecture of coping
Chirayath, G., Premamalini, K., & Joseph, J. (2025) · Frontiers in Psychology · doi:10.3389/fpsyg.2025.1699320 · Cited by 58
Metadata
- Authors
- Blease, C., & Torous, J.
- Affiliations
- Uppsala University (SE)
- Beth Israel Deaconess Medical Center (US)
- Corresponding author
- Blease, Charlotte
- Venue
- BMJ Mental Health
- Publisher
- BMJ
- Published
- 2023-11
- Volume
- 26
- Issue
- 1
- Pages
- e300884
- ISSN
- 2755-9734 (ISSN-L)
- PMC
- PMC10649440
- OpenAlex
- W4388565177
- Cited by
- 122
- References
- 12
- Citations by year
- 2026: 43 · 2025: 47 · 2024: 31 · 2023: 1
- FWCI
- 4.87
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- Locations
- BMJ Mental Health (publishedVersion, cc-by-nc)
- PubMed Central (publishedVersion, cc-by-nc)
- PubMed (publishedVersion)
- Uppsala University Publications (Uppsala University) (submittedVersion, other-oa)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- HighWire Press Open Archive (submittedVersion)
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Keywords
- ChatGPT, large language models, generative AI, mental health services, chatbots, access to mental health care, health equity, AI risks, digital mental health
- MeSH terms
- Artificial Intelligence, Communication, Health Facilities, Humans, Mental Health, Mental Health Services
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: ChatGPT and mental healthcare: balancing benefits with risks of harms. https://app.cochat.ai/papers/10.1136/bmjment-2023-300884
- The paper
- Blease, C., & Torous, J. (2023). ChatGPT and mental healthcare: balancing benefits with risks of harms. BMJ Mental Health. https://doi.org/10.1136/bmjment-2023-300884
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.