{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T17:00:02Z","timestamp":1791478802671,"version":"4.3.4"},"reference-count":28,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"<jats:p>In this Viewpoint, we highlight the principal cybersecurity measures that should be implemented to facilitate safe and effective integration of large language models (LLMs) into health care and propose a conceptual, life cycle\u2013based framework synthesizing evidence from security and clinical informatics literature. While LLMs offer significant potential for applications in clinical documentation, triage, and medical education, their deployment creates novel vulnerabilities that can compromise patient safety and data confidentiality. We argue that these vulnerabilities must be addressed across the entire deployment life cycle, with distinct threats arising before and after a model enters clinical use. Predeployment risks include data and model poisoning, where an LLM\u2019s training data or core parameters are maliciously corrupted to embed biases or backdoors. After deployment, LLMs are susceptible to inference attacks, such as prompt injection and adversarial inputs, which can be used to manipulate model behavior and extract sensitive information. Standard performance benchmarks are often insufficient to detect these sophisticated attacks. Therefore, we argue that a proactive, multilayered security framework combining technical safeguards, rigorous governance, and human-in-the-loop oversight is essential for the safe and trustworthy adoption of LLMs in clinical practice.<\/jats:p>","DOI":"10.2196\/101715","type":"journal-article","created":{"date-parts":[[2026,9,23]],"date-time":"2026-09-23T15:25:09Z","timestamp":1790177109000},"page":"e101715","source":"Crossref","is-referenced-by-count":0,"title":["Cybersecurity of Large Language Models Across the Deployment Life Cycle in Health Systems"],"prefix":"10.2196","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0009-0006-8379-9912","authenticated-orcid":false,"given":"Chun Joo","family":"Goh","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-5928-019X","authenticated-orcid":false,"given":"Haoran","family":"Liu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-4970-4424","authenticated-orcid":false,"given":"Nethum","family":"Devendra","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2834-1072","authenticated-orcid":false,"given":"Khoa Nguyen","family":"Cao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,10,8]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMe2206291"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1056\/CAT.23.0404"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1001\/jamanetworkopen.2024.8857"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1093\/postmj\/qgad137"},{"key":"ref5","unstructured":"OWASP top 10 for LLM applications 2025OWASP Foundation202411172025-08-24https:\/\/genai.owasp.org\/resource\/owasp-top-10-for-llm-applications-2025\/"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-024-03445-1"},{"key":"ref7","unstructured":"WanAWallaceEShenSKleinDPoisoning language models during instruction tuning2023ICML'23: Proceedings of the 40th International Conference on Machine Learning23-29 July 2023HonoluluNorfolk, MAJMLR.org3541325"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-024-01282-7"},{"key":"ref9","unstructured":"WolfTDebutLSanhVChaumondJDelangueCMoiACistacPRaultTLoufRFuntowiczMDavisonJShleiferSvon PlatenPMaCJerniteYPluJXuCLe ScaoTGuggerSDrameMLhoestQRushATransformers: state-of-the-art natural language processing20202020 Conference on Empirical Methods in Natural Language ProcessingNov 16-20, 2020VirtualStroudsburg, PAAssociation for Computational Linguistics3845"},{"key":"ref10","unstructured":"MengKBauDAndonianABelinkovYLocating and editing factual associations in GPT2022NIPS '22: Proceedings of the 36th International Conference on Neural Information Processing SystemsNov 28-Dec 9 2022LouisianaNew York, NYCurran Associates Inc1735972"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2401.05566"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2408.12798"},{"key":"ref13","unstructured":"CarliniNPalekaDDvijothamKSteinkeTHayaseJCooperAFLeeKJagielskiMNasrMConmyAWallaceERolnickDTram\u00e8rFStealing part of a production language model2024ICML'24: Proceedings of the 41st International Conference on Machine LearningJul 21-27, 2024ViennaNorfolk, MAJMLR.org5680705"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2211.09110"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2403.03744"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.759"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2505.08775"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-025-04151-2"},{"key":"ref19","unstructured":"WeiAHaghtalabNSteinhardtJJailbroken: how does LLM safety training fail?2023NIPS '23: Proceedings of the 37th International Conference on Neural Information Processing SystemsDec 10-16, 2023Louisiana80079110"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2302.12173"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-55631-x"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.6572"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2405.15589"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2504.03759"},{"key":"ref25","doi-asserted-by":"crossref","unstructured":"AgarwalDFabbriARisherBLabanPJotySWuCSPrompt leakage effect and mitigation strategies for multi-turn LLM applications2024Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry TrackNov 12-16, 2024MiamiStroudsburg, PAAssociation for Computational Linguistics125575","DOI":"10.18653\/v1\/2024.emnlp-industry.94"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2504.11703"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2506.09420"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2410.10760"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e101715","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T16:45:09Z","timestamp":1791477909000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e101715"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10,8]]},"references-count":28,"URL":"https:\/\/doi.org\/10.2196\/101715","relation":{"has-preprint":[{"id-type":"doi","id":"10.2196\/preprints.101715","asserted-by":"object"}]},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,8]]}}}