arXiv cs.AI· Anmol Pandey, Aditya Jain, Liang Chen, Carsten Maple·· 12 小时前AI 评分44
为何逐动作授权在 LLM 智能体中无法组合成全局授权
Local Allow, Global Deny: Why Per-Action Authorization Does Not Compose in LLM Agents
AI 导读
研究指出，LLM 智能体的逐动作授权无法组合为全局授权：跨意图、目的、来源、同意与委托的全局策略违规，可由各自单独获批的执行或其组合触发。关系型目的违规属于 2-safety 超属性，不存在单条轨迹的归责点，因此不误报的逐动作监控无法检出，所调查的六种护栏均未实现跨执行监控。
正文 · 原文
Abstract:Autonomous LLM agents act through tools under per-action authorization: a local guard screens each tool call in isolation. We show that local authorization does not compose into global authorization. A compositional authorization failure (CAF) is a locally authorized execution, or a pair of locally authorized executions, that violates a global policy spanning intent, purpose, provenance, consent, and delegation. We characterize CAFs along two axes, expressiveness (single-trace safety versus relational k-safety) and observability (a ladder of monitor profiles). Our central result is an impossibility: relational purpose violations form a 2-safety hyperproperty with no per-trace locus of blame, so a per-action monitor that never raises a false alarm detects none of them, independent of the underlying model or decision procedure; they become visible only under cross-execution monitoring, which none of the six guardrails we survey implements. We make this constructive with AL-SYNTH, a SAT/SMT synthesizer (using no language model) that emits minimal CAF witnesses, each checked against an exhaustive oracle. The empirical signature is sharp: across nine LLM judges the per-action view approves the synthesized witness almost without exception, blocking it at a rate comparable to a benign control, yet the same models flag it when shown the two executions together. Per-action authorization is therefore structurally insufficient for relational policy conjuncts not reducible to single-trace safety: to bound them, a monitor must compare executions or accept false alarms.
| Comments: | Extended version of a paper accepted at the 8th IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (IEEE TPS 2026), with full proofs and appendices. 12 pages, 5 figures, 7 tables. Code and data: this https URL |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO) |
| ACM classes: | D.4.6; F.3.1; I.2.11 |
| Cite as: | arXiv:2610.27620 [cs.CR] |
| (or arXiv:2610.27620v1 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2610.27620 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Anmol Pandey [view email]
[v1]
Tue, 20 Oct 2026 04:35:39 UTC (96 KB)
来源：arXiv cs.AI · arxiv.org