Newest papers

Local Allow, Global Deny: Why Per-Action Authorization Does Not Compose in LLM Agents

Anmol Pandey, Aditya Jain, Liang Chen, Carsten Maple · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.27620 · cited by 0 works (OpenAlex) · open-access PDF

Chinese

Perjunction Autonomous Single Authorization Agent (CAF) 是按行动授权执行的相同的警报工具:一个本地警卫单独显示每个工具呼叫。我们表明,本地授权不构成全球授权。一个组成授权失败(CAF)是本地授权的执行,或者是本地授权的执行一对,这违反了全球政策扩展的意图,目的,来源,同意和委托。我们描绘了CAF沿着两个轴,表达性(单轨安全控制与关系安全)和可观察性(监控档案的阶梯)。我们的中心结果是不可能的:关系目的违反必须形成一个2安全性超级财产,没有分歧的监控位置,所以一个比行动监测器从不提高安全性警报,从未检测到虚假警报

Machine translation (Meta M2M100); the English original follows.

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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Guard (computer science), Computer security, Constructive, Authorization, Bridging (networking)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange