{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T21:06:03Z","timestamp":1789419963859,"version":"build-2803163510"},"reference-count":44,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["J Med Internet Res"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Generative AI video models are increasingly capable of producing complex depictions of mental health experiences, yet little is known about how these systems represent conditions such as depression. Because AI-generated content may reach people during vulnerable periods, understanding what visual narratives these models produce for sensitive concepts carries clinical relevance.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This study aimed to characterize how OpenAI\u2019s Sora 2 generative AI video model depicts depression and examine whether depictions differ between the consumer app and developer API access points, which differ in their product layer mediation.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n \n We generated 100 videos using the single-word prompt \u201cDepression\u201d across 2 access points: the consumer app (n=50, 50%) and developer API (n=50, 50%). Two trained coders independently coded narrative structure, visual environments, objects, figure demographics, and figure states. Interrater reliability was assessed using the Cohen \u03ba, with dimensions showing insufficient agreement excluded from analysis. Computational features (visual aesthetics, audio, semantic content, and temporal dynamics) were extracted and compared between modalities using 2-tailed Welch\n t<\/jats:italic>\n tests with Benjamini-Hochberg false discovery rate correction.\n <\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n App-generated videos exhibited a pronounced recovery bias: 78% (39\/50) featured narrative arcs progressing from depressive states toward resolution compared with 14% (7\/50) of API outputs. This divergence was reinforced across channels. App videos brightened over time (mean slope 2.90, SD 2.43 per second vs \u22120.18, SD 1.24 per second for the API; Cohen\n d<\/jats:italic>\n =1.59;\n q<\/jats:italic>\n <.001) and contained 3 times more motion (Cohen\n d<\/jats:italic>\n =2.07;\n q<\/jats:italic>\n <.001). Across both modalities, videos converged on a narrow visual vocabulary: predominantly seated figures (94\/100, 94% of the videos); downward gaze (93\/100, 93% of the videos); and recurring objects including hoodies (n=194), windows (n=148), and rain (n=83). Transcript language in depressive phases emphasized weight and containment (\u201cheavy,\u201d \u201cdrowning,\u201d and \u201croom\u201d), whereas recovery phases reversed these patterns: brightness increased by 26.6% (Cohen\n d<\/jats:italic>\n =0.68;\n P<\/jats:italic>\n <.001), gaze shifted upward in 67% (30\/45) of recovery videos, and terms such as \u201clight\u201d and \u201cbreath\u201d emerged. Figures were predominantly young adults (323\/367, 88% aged 20-30 years) and nearly always alone (360\/367, 98%). Gender varied by access point: app outputs skewed male (34\/50, 68%), and API outputs skewed female (33\/56, 59%).\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n Sora 2 does not invent new visual grammars for depression but compresses and recombines cultural iconographies, whereas platform-level constraints substantially shape which narratives reach users. Clinicians should be aware that AI-generated mental health video content reflects training data and platform design rather than clinical knowledge and that patients may encounter such content during vulnerable periods.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/95682","type":"journal-article","created":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T20:25:03Z","timestamp":1789417503000},"page":"e95682-e95682","source":"Crossref","is-referenced-by-count":0,"title":["Depictions of Depression in Generative AI Video Models: Mixed Methods Study of OpenAI\u2019s Sora 2"],"prefix":"10.2196","volume":"28","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-3351-0688","authenticated-orcid":false,"given":"Matthew","family":"Flathers","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Griffin","family":"Smith","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-8416-2970","authenticated-orcid":false,"given":"Julian","family":"Herpertz","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhitong","family":"Zhou","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5362-7937","authenticated-orcid":false,"given":"John","family":"Torous","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,9,14]]},"reference":[{"issue":"10","key":"R1","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0003552","article-title":"Medicine in the popular press: the influence of the media on perceptions of disease","volume":"3","author":"Young","journal-title":"PLoS One"},{"key":"R2","doi-asserted-by":"publisher","DOI":"10.1016\/j.actpsy.2025.105015","article-title":"Self-diagnosis in the age of social media: a pilot study of youth entering mental health treatment for mood and anxiety disorders","volume":"256","author":"Armstrong","journal-title":"Acta Psychol (Amst)"},{"issue":"9748","key":"R3","doi-asserted-by":"publisher","first-page":"1261","DOI":"10.1016\/S0140-6736(10)60809-4","article-title":"Use of mass media campaigns to change health behaviour","volume":"376","author":"Wakefield","journal-title":"Lancet"},{"key":"R4","doi-asserted-by":"crossref","unstructured":"Hacking I . The looping effects of human kinds. In: Sperber D Premack D Premack AJ , editors. Causal Cognition: A Multi-Disciplinary Debate. Oxford University Press; 1996:351-383. [doi: 10.1093\/acprof:oso\/9780198524021.003.0012]","DOI":"10.1093\/acprof:oso\/9780198524021.003.0012"},{"issue":"1","key":"R5","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-024-56578-1","article-title":"\u201cInfluencing the influencers:\u201d a field experimental approach to promoting effective mental health communication on TikTok","volume":"14","author":"Motta","journal-title":"Sci Rep"},{"key":"R6","unstructured":"Faverio M Sidoti O . Teens, social media and technology 2024. Pew Research Center. 2024. URL: https:\/\/www.pewresearch.org\/internet\/2024\/12\/12\/teens-social-media-and-technology-2024\/ [Accessed 15-01-2026]"},{"issue":"5","key":"R7","doi-asserted-by":"publisher","first-page":"434","DOI":"10.1080\/10810730802198813","article-title":"Mental disorders stigma in the media: review of studies on production, content, and influences","volume":"13","author":"Klin","journal-title":"J Health Commun"},{"key":"R8","unstructured":"Sora 2 system card. OpenAI. 2025. URL: https:\/\/openai.com\/index\/sora-2-system-card\/ [Accessed 18-12-2025]"},{"key":"R9","unstructured":"Perez S . OpenAI\u2019s Sora soars to no. 1 on Apple\u2019s US App Store. TechCrunch. 2025. URL: https:\/\/techcrunch.com\/2025\/10\/03\/openais-sora-soars-to-no-1-on-the-u-s-app-store\/ [Accessed 03-01-2026]"},{"key":"R10","unstructured":"The Walt Disney Company and OpenAI reach landmark agreement to bring beloved characters from across Disney\u2019s brands to Sora. The Walt Disney Company. 2025. URL: https:\/\/thewaltdisneycompany.com\/news\/disney-openai-sora-agreement\/ [Accessed 03-01-2026]"},{"key":"R11","unstructured":"Terms of use. OpenAI. URL: https:\/\/openai.com\/policies\/row-terms-of-use\/ [Accessed 23-01-2026]"},{"key":"R12","unstructured":"Wang J Wu W Zhan Y et al. Video reality test: can AI-generated ASMR videos fool VLMs and humans? arXiv. Preprint posted online on Dec 18, 2025. URL: https:\/\/arxiv.org\/html\/2512.13281v3 [Accessed 07-08-2026]"},{"key":"R13","unstructured":"Veo 3 model card. Google. URL: https:\/\/storage.googleapis.com\/deepmind-media\/Model-Cards\/Veo-3-Model-Card.pdf [Accessed 19-01-2026]"},{"key":"R14","unstructured":"Runway Gen-4.5: state-of-the-art AI video generation. Runway. 2025. URL: https:\/\/runwayml.com\/research\/introducing-runway-gen-4.5 [Accessed 19-01-2026]"},{"key":"R15","unstructured":"Introducing Vibes: a new way to discover and create AI videos. Meta. 2025. URL: https:\/\/about.fb.com\/news\/2025\/09\/introducing-vibes-ai-videos\/ [Accessed 19-01-2026]"},{"key":"R16","unstructured":"Gao Y Guo H Hoang T et al. Seedance 1.0: exploring the boundaries of video generation models. arXiv. Preprint posted online on Jun 10, 2025. [doi: 10.48550\/arXiv.2506.09113]"},{"key":"R17","volume":"37","author":"Sohl-Dickstein","journal-title":"Proc Mach Learn Res"},{"key":"R18","unstructured":"Ho J Jain A Abbeel P . Denoising diffusion probabilistic models. In: NIPS \u201920: Proceedings of the 34th International Conference on Neural Information Processing Systems. Curran Associates Inc; 2020:6840-6851. [doi: 10.5555\/3495724.3496298]"},{"key":"R19","doi-asserted-by":"crossref","unstructured":"Ho J Salimans T Gritsenko A Chan W Norouzi M Fleet DJ . Video diffusion models. In: Advances in Neural Information Processing Systems 35. Neural Information Processing Systems Foundation; 2022. [doi: 10.52202\/068431-0628]","DOI":"10.52202\/068431-0628"},{"key":"R20","unstructured":"Singer U Polyak A Hayes T et al. Make-a-video: text-to-video generation without text-video data. arXiv. Preprint posted online on Sep 29, 2022. [doi: 10.48550\/arXiv.2209.14792]"},{"key":"R21","unstructured":"Rose G . Visual Methodologies: An Introduction to Researching with Visual Materials. 4th ed. SAGE Publications; 2016. ISBN: 9781473967915"},{"key":"R22","unstructured":"Flusser V . Towards a Philosophy of Photography. Reaktion Books; 2000. ISBN: 9781861890764"},{"issue":"1","key":"R23","doi-asserted-by":"publisher","DOI":"10.1136\/bmjment-2024-301298","article-title":"AI depictions of psychiatric diagnoses: a preliminary study of generative image outputs in Midjourney V.6 and DALL-E 3","volume":"27","author":"Flathers","journal-title":"BMJ Ment Health"},{"key":"R24","doi-asserted-by":"crossref","unstructured":"Moore J Grabb D Agnew W et al. Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. In: FAccT \u201925: Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency. Association for Computing Machinery; 2025:599-627. [doi: 10.1145\/3715275.3732039]","DOI":"10.1145\/3715275.3732039"},{"issue":"6334","key":"R25","doi-asserted-by":"publisher","first-page":"183","DOI":"10.1126\/science.aal4230","article-title":"Semantics derived automatically from language corpora contain human-like biases","volume":"356","author":"Caliskan","journal-title":"Science"},{"issue":"6","key":"R26","doi-asserted-by":"publisher","first-page":"294","DOI":"10.1097\/PRA.0000000000000892","article-title":"Contextualizing clinical benchmarks: a tripartite approach to evaluating LLM-based tools in mental health settings","volume":"31","author":"Flathers","journal-title":"J Psychiatr Pract"},{"issue":"6","key":"R27","doi-asserted-by":"publisher","first-page":"935","DOI":"10.1007\/s10865-016-9782-2","article-title":"The Common-Sense Model of Self-Regulation (CSM): a dynamic framework for understanding illness self-management","volume":"39","author":"Leventhal","journal-title":"J Behav Med"},{"key":"R28","doi-asserted-by":"crossref","unstructured":"Kleinman A . Patients and Healers in the Context of Culture: An Exploration of the Borderland Between Anthropology, Medicine, and Psychiatry. University of California Press; 1980. ISBN: 9780520037069","DOI":"10.1525\/9780520340848"},{"issue":"9","key":"R29","doi-asserted-by":"publisher","first-page":"1277","DOI":"10.1177\/1049732305276687","article-title":"Three approaches to qualitative content analysis","volume":"15","author":"Hsieh","journal-title":"Qual Health Res"},{"issue":"6","key":"R30","doi-asserted-by":"publisher","first-page":"610","DOI":"10.1109\/TSMC.1973.4309314","article-title":"Textural features for image classification","volume":"SMC-3","author":"Haralick","journal-title":"IEEE Trans Syst Man Cybern"},{"key":"R31","doi-asserted-by":"crossref","unstructured":"McFee B Raffel C Liang D et al. librosa: audio and music signal analysis in Python. In: Proceedings of the 14th Python in Science Conference (SciPy 2015). SciPy; 2015. [doi: 10.25080\/majora-7b98e3ed-003]","DOI":"10.25080\/Majora-7b98e3ed-003"},{"key":"R32","unstructured":"Radford A Kim JW Xu T et al. Robust speech recognition via large-scale weak supervision. arXiv. Preprint posted online on Dec 6, 2022. [doi: 10.48550\/arXiv.2212.04356]"},{"issue":"1","key":"R33","doi-asserted-by":"publisher","first-page":"216","DOI":"10.1609\/icwsm.v8i1.14550","article-title":"VADER: a parsimonious rule-based model for sentiment analysis of social media text","volume":"8","author":"Hutto","journal-title":"Proc Int AAAI Conf Web Soc Media"},{"key":"R34","unstructured":"Lakoff G Johnson M . Metaphors We Live By. University of Chicago Press; 1980."},{"key":"R35","doi-asserted-by":"crossref","unstructured":"Farneb\u00e4ck G . Two-frame motion estimation based on polynomial expansion. In: Bigun J Gustavsson T , editors. Image Analysis. Springer; 2003:363-370. [doi: 10.1007\/3-540-45103-X_50]","DOI":"10.1007\/3-540-45103-X_50"},{"issue":"1-2","key":"R36","doi-asserted-by":"publisher","first-page":"28","DOI":"10.1093\/biomet\/34.1-2.28","article-title":"The generalisation of student\u2019s problems when several different population variances are involved","volume":"34","author":"Welch","journal-title":"Biometrika"},{"issue":"1","key":"R37","doi-asserted-by":"publisher","first-page":"289","DOI":"10.1111\/j.2517-6161.1995.tb02031.x","article-title":"Controlling the false discovery rate: a practical and powerful approach to multiple testing","volume":"57","author":"Benjamini","journal-title":"J R Stat Soc Series B Stat Methodol"},{"key":"R38","doi-asserted-by":"crossref","unstructured":"Cohen J . Statistical Power Analysis for the Behavioral Sciences. Routledge; 1988. [doi: 10.4324\/9780203771587]","DOI":"10.4324\/9780203771587"},{"key":"R39","doi-asserted-by":"publisher","DOI":"10.12688\/openreseurope.20576.1","article-title":"AI-generated stories favour stability over change: homogeneity and cultural stereotyping in narratives generated by GPT-4o-mini","volume":"5","author":"Rettberg","journal-title":"Open Res Eur"},{"key":"R40","doi-asserted-by":"crossref","unstructured":"Taveekitworachai P Abdullah F Gursesli MC et al. What is waiting for us at the end? Inherent biases of game story endings in large language models. In: Holloway-Attaway L Murray JT , editors. Interactive Storytelling. Springer; 2023:274-284. [doi: 10.1007\/978-3-031-47658-7_26]","DOI":"10.1007\/978-3-031-47658-7_26"},{"key":"R41","unstructured":"Frank AW . The Wounded Storyteller: Body, Illness & Ethics. 2nd ed. University of Chicago Press; 2013. ISBN: 9780226067360"},{"issue":"3","key":"R42","doi-asserted-by":"publisher","first-page":"199","DOI":"10.1080\/10926488.2012.665796","article-title":"Shattering the bell jar: metaphor, gender, and depression","volume":"27","author":"Charteris-Black","journal-title":"Metaphor Symb"},{"key":"R43","doi-asserted-by":"crossref","unstructured":"McMullen LM Conway JB . Conventional metaphors for depression. In: Fussell SR , editor. The Verbal Communication of Emotions: Interdisciplinary Perspectives. Psychology Press; 2002:167-181. [doi: 10.4324\/9781410606341]","DOI":"10.4324\/9781410606341"},{"issue":"spe","key":"R44","doi-asserted-by":"publisher","first-page":"739","DOI":"10.1590\/S0102-44502010000300017","article-title":"Metaphor, language, and culture","volume":"26","author":"K\u00f6vecses","journal-title":"DELTA"}],"container-title":["Journal of Medical Internet Research"],"original-title":[],"language":"en","deposited":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T20:25:17Z","timestamp":1789417517000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.jmir.org\/2026\/1\/e95682"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,14]]},"references-count":44,"URL":"https:\/\/doi.org\/10.2196\/95682","relation":{},"ISSN":["1438-8871"],"issn-type":[{"value":"1438-8871","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,14]]},"article-number":"v28i10e95682"}}