对抗性攻击 LLM 规划器中的符号词汇脆弱性
Adversarially Attacking Symbolic Vocabulary Vulnerabilities In LLM Planners
AI 导读
研究者提出名为 Symbolic Swapper 的 LLM 对抗攻击框架,自动生成使用非标准符号词汇的混淆规划域,系统性削弱推理模型的规划能力。实验显示,攻击可使基础模型乃至稳健规划框架的性能下降超过 70%。分析发现,成功混淆会让模型推理质量明显退化,暴露出其对任务域的理解与实际规划能力之间的差距。该论文发表于 NeurIPS 2026。
正文 · 原文
Abstract:As large language models (LLMs) are increasingly deployed as autonomous agents, their abilities in navigating automated planning tasks have become essential. However, despite their strengths, LLMs' planning ability is not robust, particularly when they are not allowed to exploit commonsense cues in the symbolic vocabulary (e.g., names of actions, predicates, and objects) used to define planning tasks. This makes LLMs unreliable compared to symbolic planners. With recent reasoning models showing improved planning performance on tasks with obfuscated symbolic vocabularies, we utilize an adversarial attack framework to reveal to what extent the crucial weaknesses of LLM agents at joint symbolic interpretation and planning remain and whether they can reason properly when presented with severe cases. As such, our main contribution is devising an LLM-based attack model, which we call Symbolic Swapper, to automatically generate obfuscated planning domains with non-standard symbolic vocabularies that systematically break strong reasoning models' ability to plan. In doing so, we reveal how much LLMs' planning abilities can be further degraded under more advanced obfuscation schemes, allowing us to ascertain their success when having to rely on their pure planning ability. We find that attacks can decrease the planning performance of base models and even robust planning frameworks by over 70%. We further analyze the factors behind how LLM planning abilities break down and find that successful obfuscation causes models to appear to significantly degrade in reasoning quality despite them showing some inherent ability to understand the domain, revealing a gap in the perception of a task domain and being able to plan under it.
| Comments: | Published at NeurIPS 2026 |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.17910 [cs.AI] |
| (or arXiv:2611.17910v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.17910 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Stephen Obadinma [view email]
[v1]
Fri, 13 Nov 2026 08:51:07 UTC (651 KB)
来源:arXiv cs.AI · arxiv.org