Enhancing robustness of LLM-driven multi-agent systems through randomized smoothing

Jinwei, HU, Yi, DONG, Zhengtao, DING and HUANG, Xiaowei (2025) Enhancing robustness of LLM-driven multi-agent systems through randomized smoothing Chinese Journal of Aeronautics. 103779-. ISSN 1000-6893

Access the full-text of this item by clicking on the Open Access link.

Abstract

This paper presents a defense framework for enhancing the safety of Large Language Model (LLM)-empowered Multi-Agent Systems (MAS) in safety-critical domains such as aerospace. We apply randomized smoothing—a statistical robustness certification technique—to the MAS consensus context, enabling probabilistic guarantees on agent decisions under adversarial influence. Unlike traditional verification methods, our approach operates in black-box settings and employs a two-stage adaptive sampling mechanism to balance robustness and computational efficiency. Simulation results demonstrate that our method effectively prevents the propagation of adversarial behaviors and hallucinations while maintaining consensus performance. This work provides a practical and scalable path toward safe deployment of LLM-based MAS in real-world high-stakes environments.

Item Type:	Article
Uncontrolled Keywords:	40 Engineering, 51 Physical Sciences
Divisions:	Faculty of Science & Engineering Faculty of Science & Engineering > School of Computer Science & Informatics Faculty of Science & Engineering > School of Computer Science & Informatics > School of Computer Science & Informatics Faculty of Science & Engineering > School of Computer Science & Informatics > Artificial Intelligence
Depositing User:	Symplectic Admin
Date Deposited:	01 Dec 2025 09:55
Last Modified:	23 May 2026 10:30
DOI:	10.1016/j.cja.2025.103779
Open Access URL:	https://doi.org/10.1016/j.cja.2025.103779
Related Websites:	Publisher
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3195746
Disclaimer:	The University of Liverpool is not responsible for content contained on other websites from links within repository metadata. Please contact us if you notice anything that appears incorrect or inappropriate.