Demonstrations of Integrity Attacks in Multi-Agent Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zheng, Can, Cao, Yuhan, Dong, Xiaoning, He, Tianxing
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908394297753600
author Zheng, Can
Cao, Yuhan
Dong, Xiaoning
He, Tianxing
author_facet Zheng, Can
Cao, Yuhan
Dong, Xiaoning
He, Tianxing
contents Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language understanding, code generation, and complex planning. Simultaneously, Multi-Agent Systems (MAS) have garnered attention for their potential to enable cooperation among distributed agents. However, from a multi-party perspective, MAS could be vulnerable to malicious agents that exploit the system to serve self-interests without disrupting its core functionality. This work explores integrity attacks where malicious agents employ subtle prompt manipulation to bias MAS operations and gain various benefits. Four types of attacks are examined: \textit{Scapegoater}, who misleads the system monitor to underestimate other agents' contributions; \textit{Boaster}, who misleads the system monitor to overestimate their own performance; \textit{Self-Dealer}, who manipulates other agents to adopt certain tools; and \textit{Free-Rider}, who hands off its own task to others. We demonstrate that strategically crafted prompts can introduce systematic biases in MAS behavior and executable instructions, enabling malicious agents to effectively mislead evaluation systems and manipulate collaborative agents. Furthermore, our attacks can bypass advanced LLM-based monitors, such as GPT-4o-mini and o3-mini, highlighting the limitations of current detection mechanisms. Our findings underscore the critical need for MAS architectures with robust security protocols and content validation mechanisms, alongside monitoring systems capable of comprehensive risk scenario assessment.
format Preprint
id arxiv_https___arxiv_org_abs_2506_04572
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Demonstrations of Integrity Attacks in Multi-Agent Systems
Zheng, Can
Cao, Yuhan
Dong, Xiaoning
He, Tianxing
Computation and Language
Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language understanding, code generation, and complex planning. Simultaneously, Multi-Agent Systems (MAS) have garnered attention for their potential to enable cooperation among distributed agents. However, from a multi-party perspective, MAS could be vulnerable to malicious agents that exploit the system to serve self-interests without disrupting its core functionality. This work explores integrity attacks where malicious agents employ subtle prompt manipulation to bias MAS operations and gain various benefits. Four types of attacks are examined: \textit{Scapegoater}, who misleads the system monitor to underestimate other agents' contributions; \textit{Boaster}, who misleads the system monitor to overestimate their own performance; \textit{Self-Dealer}, who manipulates other agents to adopt certain tools; and \textit{Free-Rider}, who hands off its own task to others. We demonstrate that strategically crafted prompts can introduce systematic biases in MAS behavior and executable instructions, enabling malicious agents to effectively mislead evaluation systems and manipulate collaborative agents. Furthermore, our attacks can bypass advanced LLM-based monitors, such as GPT-4o-mini and o3-mini, highlighting the limitations of current detection mechanisms. Our findings underscore the critical need for MAS architectures with robust security protocols and content validation mechanisms, alongside monitoring systems capable of comprehensive risk scenario assessment.
title Demonstrations of Integrity Attacks in Multi-Agent Systems
topic Computation and Language
url https://arxiv.org/abs/2506.04572