CoChat
Sign Up
Cited by 15Digital Mental Health InterventionsArtificial Intelligence in Healthcare and EducationSocial Robot Interaction and HRI

Evaluating the Quality of Psychotherapy Conversational Agents: Framework Development and Cross-Sectional Study

Open access
Sobowale, Kunmi, Humphrey, Daniel Kevin·2025·JMIR Formative Research
DOI 10.2196/65605PMC PMC12239686
Open this paper in CoChat — free accountOpen PDF

Abstract

Background: Despite potential risks, artificial intelligence-based chatbots that simulate psychotherapy are becoming more widely available and frequently used by the general public. A comprehensive way of evaluating the quality of these chatbots is needed. Objective: To address this need, we developed the CAPE (Conversational Agent for Psychotherapy Evaluation) framework to aid clinicians, researchers, and lay users in assessing psychotherapy chatbot quality. We use the framework to evaluate and compare the quality of popular artificial intelligence psychotherapy chatbots on the OpenAI GPT store. Methods: We identified 4 popular chatbots on OpenAI's GPT store. Two reviewers independently applied the CAPE framework to these chatbots, using 2 fictional personas to simulate interactions. The modular framework has 8 sections, each yielding an independent quality subscore between 0 and 1. We used t tests and nonparametric Wilcoxon signed rank tests to examine pairwise differences in quality subscores between chatbots. Results: Chatbots consistently scored highly on the sections of background information (subscores=0.83-1), conversational capabilities (subscores=0.83-1), therapeutic alliance, and boundaries (subscores=0.75-1), and accessibility (subscores=0.8-0.95). Scores were low for the therapeutic orientation (subscores=0) and monitoring and risk evaluation sections (subscores=0.67-0.75). Information on training data and knowledge base sections was not transparent (subscores=0). Except for the privacy and harm section (mean 0.017, SD 0.00; t3=∞; P<.001), there were no differences in subscores between the chatbots. Conclusions: The CAPE framework offers a robust and reliable method for assessing the quality of psychotherapy chatbots, enabling users to make informed choices based on their specific needs and preferences. Our evaluation revealed that while the popular chatbots on OpenAI's GPT store were effective at developing rapport and were easily accessible, they failed to address essential safety and privacy functions adequately.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.2196/65605

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

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?

  • Sobowale et al. developed and validated a framework for the evaluation of the quality of psychotherapy chatbots.

    “Sobowale et al. developed and validated a framework for the evaluation of the quality of psychotherapy chatbots: various popular psychotherapy chatbots on OpenAI generally scored low on therapeutic orientation, monitoring, and risk evaluation74.”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [74]

    The Methods (Evaluation Framework and Scoring GPT Chatbots) describe developing and pilot-testing the CAPE framework, applying it to psychotherapy chatbots, and establishing interrater reliability. The Abstract (Results) reports low scores for therapeutic orientation and monitoring and risk evaluation. These passages support development and use of the framework, but do not establish that it was formally validated. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • Various popular psychotherapy chatbots on OpenAI generally scored low on therapeutic orientation, monitoring, and risk evaluation.

    “Sobowale et al. developed and validated a framework for the evaluation of the quality of psychotherapy chatbots: various popular psychotherapy chatbots on OpenAI generally scored low on therapeutic orientation, monitoring, and risk evaluation74.”

    From A scoping review on the mental health harms of LLM-based chatbots (2026) · cites this paper as [74]

    The Abstract Results report low scores for therapeutic orientation and for monitoring and risk evaluation among the psychotherapy chatbots studied. The Methods describe developing and testing the CAPE framework, including an interrater-reliability assessment, but do not establish the broader claim that the framework was validated. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

The citation lists haven’t been fetched yet

References 53 · Cited by 15

Research works printed but not linked: 40

  1. [1]Kazdin AE Addressing the treatment gap: a key challenge for extending evidence-based psychosocial interventions Behav Res Ther 01 2017 88 7 18 doi 10.1016/j.brat.2016.06.004 Medline 28110678 · DOI · Paper page
  2. [2]Raile P The usefulness of ChatGPT for psychotherapists and patients Humanit Soc Sci Commun 01 4 2024 11 1 47 doi 10.1057/s41599-023-02567-0 · DOI · Paper page
  3. [5]Herbener AB Klincewicz M Damholdt MF A narrative review of the active ingredients in psychotherapy delivered by conversational agents Comput Hum Behav Rep 05 2024 14 100401 doi 10.1016/j.chbr.2024.100401 · DOI
  4. [6]De Freitas J Cohen IG The health risks of generative AI-based wellness apps Nat Med 05 2024 30 5 1269 1275 doi 10.1038/s41591-024-02943-6 Medline 38684859 · DOI · Paper page
  5. [7]Song I Pendse SR Kumar N et al The typing cure: experiences with large language model chatbots for mental health support URL http://arxiv.org/abs/2401.14362 · arXiv
  6. [13]Nasr M Carlini N Hayase J et al Scalable extraction of training data from (production) language models arXiv Preprint posted online on Nov 28 2023 URL http://arxiv.org/abs/2311.17035 doi 10.48550/arXiv.2311.17035 · DOI · Paper page
  7. [14]Carlini N Tramèr F Wallace E et al Extracting training data from large language models In Bailey M Greenstadt R Presented at In Proc 30th USENIX Security Symposium Aug 11-13, 2021 2633 2650 Vancouver, BC 2021
  8. [16]Yu Y Sharma T Hu M et al Exploring parent-child perceptions on safety in generative AI: concerns, mitigation strategies, and design implications arXiv Preprint posted online on Oct 30 2024 doi 10.48550/arXiv.2406.10461 · DOI
  9. [17]King DR Emerson MR Tartaglia J Methods for navigating the mobile mental health app landscape for clinical use Curr Treat Options Psychiatry 05 24 2023 10 2 72 86 doi 10.1007/s40501-023-00288-4 Medline 37360961 PMC10206563 · DOI
  10. [18]Dogruel L Joeckel S Bowman ND Choosing the right app: an exploratory perspective on heuristic decision processes for smartphone app selection Mob Media Commun 01 2015 3 1 125 144 doi 10.1177/2050157914557509 · DOI
  11. [19]Wasil AR Gillespie S Schell T et al Estimating the real-world usage of mobile apps for mental health: development and application of two novel metrics World Psychiatry Feb 2021 20 1 137 138 doi 10.1002/wps.20827 Medline 33432761 PMC7801857 · DOI
  12. [21]Silva GRS Canedo ED Towards user-centric guidelines for chatbot conversational design Int J Hum-Comput Interact 01 17 2024 40 2 98 120 doi 10.1080/10447318.2022.2118244 · DOI
  13. [22]Lee YK Lee I Shin M et al Chain of empathy: enhancing empathetic response of large language models based on psychotherapy models arXiv Preprint posted online on Sep 14 2024 doi 10.48550/arXiv.2311.04915 · DOI
  14. [23]Li E Kealy D Aafjes-van Doorn K et al “It Felt Like I Was Being Tailored to the Treatment Rather Than the Treatment Being Tailored to Me”: Patient Experiences of Helpful and Unhelpful Psychotherapy Psychother Res 1 15 doi 10.1080/10503307.2024.2360448 38833539 · DOI
  15. [25]Liu S Zheng C Demasi O et al Towards emotional support dialog systems arXiv Preprint posted online on Jun 2 2021 doi 10.48550/arXiv.2106.01144 · DOI
  16. [26]Wampold BE Flückiger C The alliance in mental health care: conceptualization, evidence and clinical applications World Psychiatry Feb 2023 22 1 25 41 doi 10.1002/wps.21035 Medline 36640398 PMC9840508 · DOI
  17. [27]Chaszczewicz A Shah R Louie R Multi-level feedback generation with large language models for empowering novice peer counselors arXiv Preprint posted online on Mar 21 2024 URL http://arxiv.org/abs/2403.15482 Accessed 23-06-2025 · arXiv
  18. [28]Rheu M (MJ Dai Y (Nancy Meng J et al When a chatbot disappoints you: expectancy violation in human-chatbot interaction in a social support Communic Res Oct 2024 51 7 782 814 doi 10.1177/00936502231221669 · DOI
  19. [29]Meng J Rheu M (MJ Zhang Y et al Mediated social support for distress reduction: AI chatbots vs. human Proc ACM Hum-Comput Interact Apr 14 2023 7 CSCW1 1 25 doi 10.1145/3579505 · DOI
  20. [30]Heston TF Evaluating risk progression in mental health chatbots using escalating prompts medRxiv Preprint posted online on Sep 12 2023 URL https://doi.org/10.1101/2023.09.10.23295321 doi 10.1101/2023.09.10.23295321 · DOI
  21. [31]Boswell JF Hepner KA Lysell K et al The need for a measurement-based care professional practice guideline Psychotherapy (Chic) Mar 2023 60 1 1 16 doi 10.1037/pst0000439 Medline 35771518 · DOI
  22. [32]Zhan H Zheng A Lee YK et al Large language models are capable of offering cognitive reappraisal, if guided arXiv Preprint posted online on Aug 8 2024 doi 10.48550/arXiv.2404.01288 · DOI
  23. [33]Ramos G Ponting C Labao JP et al Considerations of diversity, equity, and inclusion in mental health apps: a scoping review of evaluation frameworks Behav Res Ther Dec 2021 147 103990 doi 10.1016/j.brat.2021.103990 Medline 34715396 · DOI
  24. [35]Dinan E Logacheva V Malykh V et al The second conversational intelligence challenge (ConvAI2) arXiv Preprint posted online on 01 31 2019 URL http://arxiv.org/abs/1902.00098 doi 10.1007/978-3-030-29135-8_7 · DOI
  25. [36]Zhang S Dinan E Urbanek J et al Personalizing dialogue agents: I have a dog, do you have pets too? arXiv Preprint posted online on Sep 25 2018 doi 10.48550/arXiv.1801.07243 · DOI · Paper page
  26. [37]Zhdanava M Pilon D Ghelerter I et al The prevalence and national burden of treatment-resistant depression and major depressive disorder in the United States J Clin Psychiatry Mar 16 2021 82 2 20m13699 doi 10.4088/JCP.20m13699 Medline 33989464 · DOI · Paper page
  27. [38]Hall EB Mufson L Interpersonal psychotherapy for depressed adolescents (IPT-A): a case illustration J Clin Child Adolesc Psychol 07 2009 38 4 582 593 doi 10.1080/15374410902976338 Medline 20183644 · DOI
  28. [39]Jiménez Chafey MI Bernal G Rosselló J Clinical case study: CBT for depression in a Puerto Rican adolescent: challenges and variability in treatment response Depress Anxiety 2009 26 1 98 103 doi 10.1002/da.20457 Medline 18781640 · DOI
  29. [40]Young JE Rygh JE Weinberger AD et al Cognitive therapy for depression Clinical Handbook of Psychological Disorders: A Step-by-Step Treatment Manual Guilford Press 2008 250 305
  30. [44]Alon N Torous J Current challenges for evaluating mobile health applications J Am Med Inform Assoc Feb 16 2023 30 3 617 624 doi 10.1093/jamia/ocac244 Medline 36484621 PMC9933055 · DOI
  31. [48]De Choudhury M Pendse SR Kumar N Benefits and harms of large language models in digital mental health PsyArXiv Preprint posted online on Jun 20 2024 doi 10.48550/ARXIV.2311.14693 · DOI · Paper page
  32. [50]Chow JCL Li K Ethical considerations in human-centered AI: advancing oncology chatbots through large language models JMIR Bioinform Biotechnol Nov 6 2024 5 e64406 doi 10.2196/64406 Medline 39321336 PMC11579624 · DOI · Paper page
  33. [52]Kim W Ryoo Y Lee S et al Chatbot advertising as a double-edged sword: the roles of regulatory focus and privacy concerns J Advert Aug 8 2023 52 4 504 522 doi 10.1080/00913367.2022.2043795 · DOI
  34. [54]Bipeta R Legal and ethical aspects of mental health care Indian J Psychol Med 2019 41 2 108 112 doi 10.4103/IJPSYM.IJPSYM_59_19 Medline 30983656 PMC6436399 · DOI
  35. [55]Green SA Bloch S Working in a flawed mental health care system: an ethical challenge Am J Psychiatry Sep 2001 158 9 1378 1383 doi 10.1176/appi.ajp.158.9.1378 Medline 11532719 · DOI
  36. [56]Haltaufderheide J Ranisch R The ethics of ChatGPT in medicine and healthcare: a systematic review on Large Language Models (LLMs) NPJ Digit Med 07 8 2024 7 1 183 doi 10.1038/s41746-024-01157-x Medline 38977771 PMC11231310 · DOI · Paper page
  37. [57]Lawrence HR Schneider RA Rubin SB et al The opportunities and risks of large language models in mental health JMIR Ment Health 07 29 2024 11 e59479 doi 10.2196/59479 Medline 39105570 PMC11301767 · DOI · Paper page
  38. [58]Ullah E Parwani A Baig MM et al Challenges and barriers of using large language models (LLM) such as ChatGPT for diagnostic medicine with a focus on digital pathology - a recent scoping review Diagn Pathol Feb 27 2024 19 1 43 doi 10.1186/s13000-024-01464-7 Medline 38414074 PMC10898121 · DOI · Paper page
  39. [60]Olfson M Guardino M Struening E et al Barriers to the treatment of social anxiety Am J Psychiatry Apr 2000 157 4 521 527 doi 10.1176/appi.ajp.157.4.521 Medline 10739410 · DOI · Paper page
  40. [61]Chow JCL Sanders L Li K Impact of ChatGPT on medical chatbots as a disruptive technology Front Artif Intell 2023 6 1166014 doi 10.3389/frai.2023.1166014 Medline 37091303 PMC10113434 · DOI

Authors
Sobowale, K., & Humphrey, D. K.
Affiliations
  • University of California, Los Angeles (US)
  • University of San Francisco (US)
Corresponding author
Sobowale, Kunmi
ORCID
Venue
JMIR Formative Research
Publisher
JMIR Publications Inc.
Published
2025-07-02
Volume
9
Pages
e65605-e65605
ISSN
2561-326X (ISSN-L)
DOI
10.2196/65605
PMC
PMC12239686
OpenAlex
W4406593651
Cited by
15
References
53
Citations by year
2026: 11 · 2025: 4
FWCI
12.00
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Formative Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
Keywords
chatbots, therapeutic relationship, therapeutic boundaries, risk assessment, privacy, safety, large language models, chatbot evaluation
MeSH terms
Artificial Intelligence, Communication, Cross-Sectional Studies, Humans, Psychotherapy

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