# BAPLe++: Cross-Model Transferability and Prompt-Level Defense of Backdoors in Medical Foundation Models

DevFeed: [BAPLe++: Cross-Model Transferability and Prompt-Level Defense of Backdoors in Medical Foundation Models](<https://devfeed.tech/articles/baple-cross-model-transferability-and-prompt-level-defense-of-backdoors-in-medical-foundation-models-125250.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.10764>)

Author: Ufaq Khan, Umair Nawaz, Satyajit Tourani, Awais Rauf, Fahad Shamshad, Asif Hanif, Rao Muhammad Anwer, Muhammad Haris Khan, Mohammad Yaqub

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [few-shot](<https://devfeed.tech/topics/few-shot.md>), [prompt](<https://devfeed.tech/topics/prompt.md>), [Enhancing AI model performance](<https://devfeed.tech/topics/enhancing-ai-model-performance.md>), [Medical imaging](<https://devfeed.tech/topics/medical-imaging.md>)

Tags: [backdoor](<https://devfeed.tech/tags/backdoor.md>), [backdoors](<https://devfeed.tech/tags/backdoors.md>), [cad](<https://devfeed.tech/tags/cad.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [defense](<https://devfeed.tech/tags/defense.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [few-shot](<https://devfeed.tech/tags/few-shot.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [input](<https://devfeed.tech/tags/input.md>), [learning](<https://devfeed.tech/tags/learning.md>), [medical](<https://devfeed.tech/tags/medical.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [noise](<https://devfeed.tech/tags/noise.md>), [performance](<https://devfeed.tech/tags/performance.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [training](<https://devfeed.tech/tags/training.md>)

## AI overview

BAPLe++ studies whether backdoors implanted in prompts for medical foundation models transfer across models and whether they can be repaired without changing the backbone. The authors report asymmetric, dataset-dependent transfer across four models and six chest X-ray and histopathology datasets, measured with Clean Accuracy Difference (CAD) and Backdoor Accuracy Difference (BAD). Their DePLe method anchors on the infected prompt and learns a new prompt from a small clean set, aiming to preserve the task while moving away from the compromised embedding region. Against four training-free input transformations, DePLe reduced backdoor accuracy more consistently while preserving clean accuracy, and sometimes improving it.

## Source excerpt

arXiv:2611.10764v1 Announce Type: cross Abstract: Medical foundation models (Med-FMs) are increasingly adapted to clinical tasks through prompt learning, which tunes a small set of prompt tokens while the backbone remains frozen. Recent work has shown that this efficiency introduces a vulnerability: jointly optimizing learnable prompts and an imperceptible noise trigger during few-shot adaptation implants a backdoor that leaves clean performance intact but steers triggered inputs to an attacker-chosen class. Since prompts are compact and readily shared, this raises two questions: whether a backdoored prompt remains effective on a different Med-FM, and whether it can be repaired without retraining the backbone. We address both in BAPLe++. First, we present the first study of cross-model transferability of prompt-based backdoors, quantified by two proposed metrics, Clean Accuracy Difference (CAD) and Backdoor Accuracy Difference (BAD). Second, we propose DePLe (Defense via Prompt Learning), which uses the infected prompt as a fixed anchor and learns a new prompt on a small clean set that preserves the clean task while diverging from the compromised embedding region. Experiments on four Med-FMs and six chest X-ray and histopathology datasets show that backdoors transfer across models, but asymmetrically and in a dataset-dependent manner. Compared with four training-free input transformations, DePLe reduces backdoor accuracy more consistently while preserving, and in several settings improving, clean accuracy. Our code can be accessed on https://github.com/Ufaqkhan/Baple-Plus-Plus