About The Study: In this quality improvement study using a controlled simulation, commercial large language models (LLM’s) demonstrated substantial vulnerability to prompt-injection attacks (i.e., maliciously crafted inputs that manipulate an LLM’s behavior) that could generate clinically dangerous recommendations; even flagship models with advanced safety mechanisms showed high susceptibility. These findings underscore the need for adversarial robustness testing, system-level safeguards, and regulatory oversight before clinical deployment.
Corresponding Author: To contact the corresponding author, Jungyo Suh, MD, email uro_jun@amc.seoul.kr .
To access the embargoed study: Visit our For The Media website at this link https://media.jamanetwork.com/
(doi:10.1001/jamanetworkopen.2025.49963)
Editor’s Note: Please see the article for additional information, including other authors, author contributions and affiliations, conflict of interest and financial disclosures, and funding and support.
# # #
Embed this link to provide your readers free access to the full-text article This link will be live at the embargo time https://jamanetwork.com/journals/jamanetworkopen/fullarticle/10.1001/jamanetworkopen.2025.49963?guestAccessKey=1b34668e-afe8-4888-aa3d-dd05b3b83eff&utm_source=for_the_media&utm_medium=referral&utm_campaign=ftm_links&utm_content=tfl&utm_term=121925
About JAMA Network Open: JAMA Network Open is an online-only open access general medical journal from the JAMA Network. On weekdays, the journal publishes peer-reviewed clinical research and commentary in more than 40 medical and health subject areas. Every article is free online from the day of publication.
JAMA Network Open