新的视觉提示注入攻击针对前沿视觉语言模型

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-07 04:00

研究人员开发了一种名为Repeat-After-Me的新型黑盒自适应视觉提示注入攻击，能够提取个人身份信息或执行恶意工具调用。该方法在开放权重和商业前沿视觉语言模型（VLMs）上均取得了高成功率，包括Qwen3.6-27B和GPT-5.5。该攻击在模型之间表现出显著的迁移性，并在真实的OpenClaw代理部署中成功测试，其中注入的图像可以覆盖关键URL，可能导致远程代码执行和秘密泄露。 AI

影响这项研究突显了视觉提示注入中的关键漏洞，可能影响AI代理的安全性，并需要新的防御机制。

排序理由该集群基于一篇详细介绍针对AI模型的新攻击方法的论文。[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: 32 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群基于一篇详细介绍针对AI模型的新攻击方法的论文。[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: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Sizhe Chen, Yu-Lin Tsai, Ivan Evtimov, Kamalika Chaudhuri, Raluca Ada Popa, David Wagner, Arman Zharmagambetov · 2026-09-07 04:00

Repeat-After-Me：黑盒自适应视觉提示注入

arXiv:2609.04533v1 Announce Type: cross Abstract: Prompt injection is widely recognized as a major security threat to AI agents that interact with untrusted external data, such as websites, documents, and emails. Prior work has shown that, in the text domain, black-box prompt inj…

本文如何被排名

报道来源 [1]

Repeat-After-Me：黑盒自适应视觉提示注入

相关实体

相关话题