Position: AI Security Policy Should Target Systems, Not Models

Abstract

We present swarm-attack, an open-source adversarial testing framework in which multiple lightweight LLM agents coordinate through shared memory, parallel exploration, and evolutionary optimization. Together, our results demonstrate that both safety bypass of frontier models and software vulnerability discovery, i.e., the capability class that motivated restricted release of Anthropic's Mythos Preview, are achievable at effectively zero cost using commodity hardware and openly available models. W

Categories

Framework mappings

Suggested from the entry's categories.

@misc{riegler2026position,
  title = {{Position: AI Security Policy Should Target Systems, Not Models}},
  author = {Michael A. Riegler and Inga Strümke},
  year = {2026},
  month = may,
  eprint = {2605.09504},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.09504}
}