Adversarial robustness of LLM-based multi-agent systems for engineering problems

Large language models (LLMs) are increasingly deployed in multi-agent systems (MAS), including for solving engineering problems. Unlike purely linguistic tasks, engineering workflows demand formal rigor and numerical accuracy, meaning that adversarial perturbations can cause not just degraded performance but systematically incorrect or unsafe results. In this work, we present one of the first systematic studies of adversarial robustness of LLM-based MAS in engineering contexts. Using representative problems—including pipe pressure loss (Darcy-Weisbach), beam deflection, mathematical modeling, and graph traversal—we investigate how misleading agents affect collaborative reasoning and quantify error propagation under controlled adversarial influence. Our results show that adversarial vulnerabilities in engineering differ from those observed in generic MAS evaluations in important aspects: system robustness is sensitive to task type, the subtlety of injected errors, and communication order among agents. In particular, engineering tasks with higher structural complexity or easily confusable numerical variations are especially prone to adversarial influence. We further identify design choices, such as prompt framing, agent role assignment, and discussion order, that significantly improve resilience. These findings highlight the need for domain-specific evaluation of adversarial robustness and provide actionable insights for designing MAS that are trustworthy and safe in engineering applications.

Subjects

adversarial robustness

alignment

engineering

large language model (LLM)

misalignment

multi-agent system (MAS)

DDC Class

006.3: Artificial Intelligence

004: Computer Sciences

519: Applied Mathematics, Probabilities

Lizenz

https://creativecommons.org/licenses/by/4.0/

Publication version

publishedVersion

Name

frai-09-1784484-g0005.tif

Size

426.39 KB

Format

TIFF

Name

frai-09-1784484-g0003.tif

Size

576.45 KB

Format

TIFF

Name

frai-09-1784484-g0007.tif

Size

404.74 KB

Format

TIFF

Name

frai-09-1784484-g0004.tif

Size

719.7 KB

Format

TIFF

Name

Supplementary_file_1.pdf

Size

1.88 MB

Format

Adobe PDF

Name

frai-09-1784484-g0001.tif

Size

1.23 MB

Format

TIFF

Name

frai-09-1784484-g0002.tif

Size

405.84 KB

Format

TIFF

Name

frai-09-1784484-g0006.tif

Size

372.67 KB

Format

TIFF

Name

frai-09-1784484.pdf

Size

1.57 MB

Format

Adobe PDF

Options

Adversarial robustness of LLM-based multi-agent systems for engineering problems