研究人员开发了ARMOR++，一个新颖的多智能体框架，旨在增强针对深度伪造检测器的攻击的可迁移性。该系统利用Qwen2.5-VL视觉语言模型作为语义先验，并使用Qwen3大型语言模型来编排攻击原语。在AADD-2025基准测试上的评估表明，ARMOR++的性能显著优于现有方法，凸显了当前深度伪造检测技术中存在的持续性漏洞。 AI
影响 这项研究揭示了深度伪造检测器可靠性方面存在的重大差距，并强调了智能体编排在揭示潜在漏洞方面的潜力。
排序理由 该集群包含一篇学术论文，详细介绍了一种用于深度伪造检测器对抗性攻击的新框架。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2607.15246v1 Announce Type: new Abstract: The reliability of deepfake detectors frequently degrades under black-box adversarial transfer, as these models often rely on fragile, architecture-dependent forensic cues. Existing transfer attacks often lack semantic awareness and…
The reliability of deepfake detectors frequently degrades under black-box adversarial transfer, as these models often rely on fragile, architecture-dependent forensic cues. Existing transfer attacks often lack semantic awareness and struggle to maintain effectiveness under strict…