Saved in:
Bibliographic Details
Main Author: La Torre, Alejandro Paredes
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.16960
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912977226039296
author La Torre, Alejandro Paredes
author_facet La Torre, Alejandro Paredes
contents We study adversarial robustness of open-source vision-language model (VLM) agents deployed in a self-contained e-commerce environment built to simulate realistic pre-deployment conditions. We evaluate two agents, LLaVA-v1.5-7B and Qwen2.5-VL-7B, under three gradient-based attacks: the Basic Iterative Method (BIM), Projected Gradient Descent (PGD), and a CLIP-based spectral attack. Against LLaVA, all three attacks achieve substantial attack success rates (52.6%, 53.8%, and 66.9% respectively), demonstrating that simple gradient-based methods pose a practical threat to open-source VLM agents. Qwen2.5-VL proves significantly more robust across all attacks (6.5%, 7.7%, and 15.5%), suggesting meaningful architectural differences in adversarial resilience between open-source VLM families. These findings have direct implications for the security evaluation of VLM agents prior to commercial deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2603_16960
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Adversarial attacks against Modern Vision-Language Models
La Torre, Alejandro Paredes
Cryptography and Security
Artificial Intelligence
We study adversarial robustness of open-source vision-language model (VLM) agents deployed in a self-contained e-commerce environment built to simulate realistic pre-deployment conditions. We evaluate two agents, LLaVA-v1.5-7B and Qwen2.5-VL-7B, under three gradient-based attacks: the Basic Iterative Method (BIM), Projected Gradient Descent (PGD), and a CLIP-based spectral attack. Against LLaVA, all three attacks achieve substantial attack success rates (52.6%, 53.8%, and 66.9% respectively), demonstrating that simple gradient-based methods pose a practical threat to open-source VLM agents. Qwen2.5-VL proves significantly more robust across all attacks (6.5%, 7.7%, and 15.5%), suggesting meaningful architectural differences in adversarial resilience between open-source VLM families. These findings have direct implications for the security evaluation of VLM agents prior to commercial deployment.
title Adversarial attacks against Modern Vision-Language Models
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2603.16960