跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2610.15728 · v2

When Watermarks Turn Adversarial: Exposing Vulnerabilities in Document Understanding

首次提交 2026-10-12 00:05
版本更新 2026-11-17 01:33
雷达观测 2026-11-17 16:10
arXiv ↗PDF ↗

摘要

Visible watermarks are common in digital documents and persist during vision-language model (VLM) processing. We show that they can also serve as adversarial surfaces for document understanding. We propose Watermark-based Adversarial Surface Perturbation (WASP), a white-box attack that optimizes pixels inside fixed watermark strokes without requiring downstream questions, answers, or task-specific losses. We further introduce Balanced Aggregation Loss (BAL), which emphasizes visual token groups that remain highly similar to clean representations. Adversarial watermarks degrade performance substantially more than ordinary watermarks. On DocVQA, WASP achieves attack success rates of 96.81% and 71.93% against Qwen2.5-VL and InternVL2.5; BAL improves over a standard token hinge objective by 9.21 and 2.24 percentage points. On 1,651 OmniDocBench pages, WASP increases MinerU2.5's text edit distance by 0.7748 relative to ordinary watermarks. WASP also transfers within model families and across readers sharing a vision-tower architecture. Attack effectiveness depends on watermark geometry, model architecture, and preprocessing; in structured parsing, disrupted layout and routing can propagate errors downstream. These results expose an underexplored security risk in document understanding systems.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-17 01:33 · 最新观测版本

When Watermarks Turn Adversarial: Exposing Vulnerabilities in Document Understanding

Visible watermarks are common in digital documents and persist during vision-language model (VLM) processing. We show that they can also serve as adversarial surfaces for document understanding. We propose Watermark-based Adversarial Surface Perturbation (WASP), a white-box attack that optimizes pixels inside fixed watermark strokes without requiring downstream questions, answers, or task-specific losses. We further introduce Balanced Aggregation Loss (BAL), which emphasizes visual token groups that remain highly similar to clean representations. Adversarial watermarks degrade performance substantially more than ordinary watermarks. On DocVQA, WASP achieves attack success rates of 96.81% and 71.93% against Qwen2.5-VL and InternVL2.5; BAL improves over a standard token hinge objective by 9.21 and 2.24 percentage points. On 1,651 OmniDocBench pages, WASP increases MinerU2.5's text edit distance by 0.7748 relative to ordinary watermarks. WASP also transfers within model families and across readers sharing a vision-tower architecture. Attack effectiveness depends on watermark geometry, model architecture, and preprocessing; in structured parsing, disrupted layout and routing can propagate errors downstream. These results expose an underexplored security risk in document understanding systems.

查看此版本解读 →查看此版本原文 ↗