Guardado en:
Detalles Bibliográficos
Autores principales: Zhang, Chi, Ding, Wenxuan, Liu, Jiale, Wu, Mingrui, Wu, Qingyun, Mooney, Ray
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2601.19202
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910001938825216
author Zhang, Chi
Ding, Wenxuan
Liu, Jiale
Wu, Mingrui
Wu, Qingyun
Mooney, Ray
author_facet Zhang, Chi
Ding, Wenxuan
Liu, Jiale
Wu, Mingrui
Wu, Qingyun
Mooney, Ray
contents Vision-Language Models (VLMs) have shown strong multimodal reasoning capabilities on Visual-Question-Answering (VQA) benchmarks. However, their robustness against textual misinformation remains under-explored. While existing research has studied the effect of misinformation in text-only domains, it is not clear how VLMs arbitrate between contradictory information from different modalities. To bridge the gap, we first propose the CONTEXT-VQA (i.e., Conflicting Text) dataset, consisting of image-question pairs together with systematically generated persuasive prompts that deliberately conflict with visual evidence. Then, a thorough evaluation framework is designed and executed to benchmark the susceptibility of various models to these conflicting multimodal inputs. Comprehensive experiments over 11 state-of-the-art VLMs reveal that these models are indeed vulnerable to misleading textual prompts, often overriding clear visual evidence in favor of the conflicting text, and show an average performance drop of over 48.2% after only one round of persuasive conversation. Our findings highlight a critical limitation in current VLMs and underscore the need for improved robustness against textual manipulation.
format Preprint
id arxiv_https___arxiv_org_abs_2601_19202
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Do Images Speak Louder than Words? Investigating the Effect of Textual Misinformation in VLMs
Zhang, Chi
Ding, Wenxuan
Liu, Jiale
Wu, Mingrui
Wu, Qingyun
Mooney, Ray
Computation and Language
Vision-Language Models (VLMs) have shown strong multimodal reasoning capabilities on Visual-Question-Answering (VQA) benchmarks. However, their robustness against textual misinformation remains under-explored. While existing research has studied the effect of misinformation in text-only domains, it is not clear how VLMs arbitrate between contradictory information from different modalities. To bridge the gap, we first propose the CONTEXT-VQA (i.e., Conflicting Text) dataset, consisting of image-question pairs together with systematically generated persuasive prompts that deliberately conflict with visual evidence. Then, a thorough evaluation framework is designed and executed to benchmark the susceptibility of various models to these conflicting multimodal inputs. Comprehensive experiments over 11 state-of-the-art VLMs reveal that these models are indeed vulnerable to misleading textual prompts, often overriding clear visual evidence in favor of the conflicting text, and show an average performance drop of over 48.2% after only one round of persuasive conversation. Our findings highlight a critical limitation in current VLMs and underscore the need for improved robustness against textual manipulation.
title Do Images Speak Louder than Words? Investigating the Effect of Textual Misinformation in VLMs
topic Computation and Language
url https://arxiv.org/abs/2601.19202