Journal Article

DKFZ-2025-00276

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Prompt injection attacks on vision language models in oncology.

Clusmann, J. ; Ferber, D. ; Wiest, I. C. ; Schneider, C. V. ; Brinker, T. J.DKFZ* ; Foersch, S. ; Truhn, D. ; Kather, J. N.

2025
Springer Nature [London]

This record in other databases:  

Please use a persistent id in citations: doi:10.1038/s41467-024-55631-x

Abstract: Vision-language artificial intelligence models (VLMs) possess medical knowledge and can be employed in healthcare in numerous ways, including as image interpreters, virtual scribes, and general decision support systems. However, here, we demonstrate that current VLMs applied to medical tasks exhibit a fundamental security flaw: they can be compromised by prompt injection attacks. These can be used to output harmful information just by interacting with the VLM, without any access to its parameters. We perform a quantitative study to evaluate the vulnerabilities to these attacks in four state of the art VLMs: Claude-3 Opus, Claude-3.5 Sonnet, Reka Core, and GPT-4o. Using a set of N = 594 attacks, we show that all of these models are susceptible. Specifically, we show that embedding sub-visual prompts in manifold medical imaging data can cause the model to provide harmful output, and that these prompts are non-obvious to human observers. Thus, our study demonstrates a key vulnerability in medical VLMs which should be mitigated before widespread clinical adoption.

Keyword(s): Humans (MeSH) ; Artificial Intelligence (MeSH) ; Medical Oncology: methods (MeSH) ; Algorithms (MeSH)

---

Contributing Institute(s):

1. Digitale Prävention, Diagnostik und Therapiesteuerung (C140)

Research Program(s):

1. 313 - Krebsrisikofaktoren und Prävention (POF4-313) (POF4-313)

Appears in the scientific report 2025

---

Database coverage:
; ; ; Article Processing Charges ; BIOSIS Previews ; Biological Abstracts ; Clarivate Analytics Master Journal List ; Current Contents - Agriculture, Biology and Environmental Sciences ; Current Contents - Life Sciences ; Current Contents - Physical, Chemical and Earth Sciences ; DOAJ Seal ; Essential Science Indicators ; Fees ; IF >= 15 ; JCR ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection ; Zoological Record

---