Separation of Duties for Privileged LLM Agents: A Governed Execution Architecture with Measured Security-Utility Trade-offs

arXiv:2609.38224 · cs.CR · Submitted 2026-09-28 · Read on arXiv

cs.CR

Submitted: 2026-09-28

Updated: 2026-09-28

Comments: 12 pages, 2 figures. Supplementary material and a reproducibility artifact are provided as ancillary files

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmark — 26× in this paper · explained in 2 episodes
• Verification — 13× in this paper · explained in 2 episodes
• Failure Modes — 6× in this paper · explained in 3 episodes
• Ablation — 5× in this paper · explained in 1 episode
• Aggregation — 4× in this paper
• Benchmarking — 4× in this paper · explained in 4 episodes
• Agentic System — 3× in this paper · explained in 2 episodes
• Alignment — 3× in this paper
• In Network and Distributed System Security — 3× in this paper
• AI Agents — 2× in this paper · explained in 2 episodes
• Agentic AI — 2× in this paper · explained in 2 episodes
• Association for Computational Linguistics — 2× in this paper
• Trace — 2× in this paper · explained in 2 episodes
• Abstraction — 1× in this paper · explained in 1 episode
• Accountability — 1× in this paper · explained in 2 episodes
• Agent Security Bench — 1× in this paper
• Artificial Intelligence Security Verification Standard — 1× in this paper
• Decoupling — 1× in this paper · explained in 2 episodes
• Multi-Agent Framework — 1× in this paper · explained in 3 episodes

Sources

• A Framework for Formalizing LLM Agent Security
• Defeating Prompt Injections by Design
• Governance-as-a-Service: A Multi-Agent Framework for AI System Compliance and Policy Enforcement
• Progent: Securing AI Agents with Privilege Control
• Design Patterns for Securing LLM Agents against Prompt Injections

Related papers

• SoK: AI-Augmented Binary Reversing
• Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
• Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
• Estimating the Social Cost of Corporate Data Breaches
• Efficient Fuzzy PSI under One-Sided Assumptions
• Sealing the Audit-Runtime Gap for LLM Skills