Smarter Saboteurs, Better Fixers: Scaling & Security in Linear Multi-Agent Workflows

Abstract

As LLM-based multi-agent systems (MAS) are deployed in the wild, the resilience of their collaboration structures against adversarial compromise becomes a critical safety concern. Attackers may leverage prompt-injection or jailbreaking to sabotage individual agents within MAS workflows, but the interaction between model scaling and system-level resilience remains poorly understood. This paper investigates how model scale affects the security of linear multi-agent workflows. Our experiments acros

Categories

Framework mappings

Suggested from the entry's categories.

@misc{mcallister2026smarter,
  title = {{Smarter Saboteurs, Better Fixers: Scaling \& Security in Linear Multi-Agent Workflows}},
  author = {Timothy McAllister and Sina Abdidizaji and Ivan Garibay and Ozlem Ozmen Garibay},
  year = {2026},
  month = jun,
  eprint = {2606.12709},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.12709}
}