研究人员发现了一种在大型语言模型（LLM）中嵌入隐藏消息的新颖方法，该方法可以绕过传统的基于文本的安全措施。一种技术涉及将有效载荷作为结构化浮点参数进行传输，即使存在文本分类器也能逃避检测。另一种方法利用LLM推理中使用的伪随机数生成器，将消息嵌入到种子中，从而仅凭生成的文本就可以重建秘密。此外，一项研究表明，即使是旨在检测这些隐藏消息的复杂的内部激活探测也可以被规避，尽管特定的数据级干预可以恢复可检测性。 AI
影响 揭示了LLM安全的新攻击向量，并强调需要超越简单文本分析的更强大的检测机制。
排序理由 多篇研究论文详细介绍了LLM内的隐写术新方法及其防御措施。
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
arXiv:2606.08403v1 Announce Type: cross Abstract: Text-centered prompt-injection defenses assume that the malicious signal is visible in one of the inspected text views. We study a reproducible LLM01-style indirect prompt/content-injection failure mode where that assumption break…
arXiv:2606.09135v1 Announce Type: cross Abstract: We demonstrate that widely deployed Large Language Model (LLM) inference stacks harbor a steganographic channel that requires no modification to model weights, sampling code, or output distributions. The channel exploits a structu…
arXiv:2606.09411v1 Announce Type: cross Abstract: Large language models can be fine-tuned to encode prompt-borne secrets into fluent, seemingly benign outputs. This creates a steganographic exfiltration risk that is difficult to detect with output-level steganalysis. Recent work …
Large language models can be fine-tuned to encode prompt-borne secrets into fluent, seemingly benign outputs. This creates a steganographic exfiltration risk that is difficult to detect with output-level steganalysis. Recent work proposes mechanistic detection using linear probes…
Text-centered prompt-injection defenses assume that the malicious signal is visible in one of the inspected text views. We study a reproducible LLM01-style indirect prompt/content-injection failure mode where that assumption breaks: a payload caught in plain English slips past th…