English(EN) Adversarial Attacks on Deep OCR Systems

新的黑盒攻击针对DeepSeek-OCR，揭示了解码器故障

作者 PulseAugur 编辑部 · [1 个来源] · 2026-08-11 04:00

研究人员开发了一种新颖的黑盒对抗性攻击，针对生成式OCR视觉语言模型DeepSeek-OCR。该攻击仅基于解码后的字符串输出进行操作，无需访问模型内部的梯度或logits等信息。该方法将攻击视为一个零阶优化问题，使用随机方向有限差分方案来估计梯度，并使用带有ell_infinity投影的Adam优化器来实现不可感知的图像扰动。对DeepSeek-OCR的初步实验揭示了显著的定性解码器故障，包括重复、截断和提示泄露，尽管有针对性的重写更具挑战性。 AI

影响这项研究突显了先进OCR系统潜在的安全漏洞，促使对鲁棒性防御进行进一步研究。

排序理由详细介绍新对抗性攻击方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍新对抗性攻击方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 58 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Wenbo Sun, Hongzong LI, Yanyun Wang, Jiahao MA, Shuxin Zhuang, Rong Feng, Shiqin Tang, Zi Liang · 2026-08-11 04:00

深度OCR系统上的对抗性攻击

arXiv:2608.07636v1 Announce Type: cross Abstract: Deep-OCR (DeepSeek-OCR) advances document recognition by treating the visual modality as an optical compression medium, enabling long-context OCR at low token cost. However, its increased complexity may introduce new security vuln…

本文如何被排名

报道来源 [1]

深度OCR系统上的对抗性攻击

相关实体

相关话题