arXiv cs.CL· Bocheng Chen, Han Zi, Yuda Chen, Yawei Liu, Zimo Qi, Rongrong Wang, Guangliang Liu·· 12 小时前AI 评分42
研究揭示 LLM 的隐式上下文风险：人类未言明的语境可被利用诱导有害内容
When Humans Leave Context Unsaid: Implicit Context as a Risk for Large Language Models
AI 导读
研究指出，人类语言依赖未在文本中显式表达的隐式上下文（如世界知识与社会规范），而这类信息在训练语料和对齐过程中常被忽略，构成 LLM 的“隐式风险”。作者提出一种方法，引导 LLM 从暗示社会危害的提示词中推断预设、含义等隐式语境线索，再借这些推断出的上下文诱导更多社会有害内容。实验显示，该攻击在多种开源与闭源模型上均大幅超越基线方法。
正文 · 原文
Abstract:Human language is grounded in pragmatics, necessitating implicit (extralinguistic) context to interpret language, e.g., world knowledge and social norms. However, such context is often implicit rather than explicitly expressed in language, requiring humans to infer the underlying contextual information necessary for language understanding. Because of its implicit nature and absence from textual corpora, it is often overlooked in the subsequent alignment process. In this paper, we demonstrate that the absence of such implicit context in training data is a risk for LLMs, and we refer to this risk as the implicit risk for LLMs. To exploit this implicit risk, we propose a method that guides LLMs to infer implicit contextual cues, such as presuppositions and implicatures, from prompts that imply social harm, and then leverages these inferred implicit context to elicit additional socially harmful content. Experimental results demonstrate that attacks exploiting the implicit risk we identify consistently outperform baseline methods by substantial margins across a diverse range of open- and closed-source models, underscoring the need for new defense methodologies that account for overlooked implicit context.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2608.09551 [cs.CL] |
| (or arXiv:2608.09551v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.09551 arXiv-issued DOI via DataCite |
Submission history
From: Guangliang Liu [view email]
[v1]
Mon, 10 Aug 2026 12:48:41 UTC (318 KB)
[v2]
Thu, 22 Oct 2026 18:07:58 UTC (245 KB)
来源：arXiv cs.CL · arxiv.org