HoloAegis: Frozen Representation, Topological Inference: Minimally Parametric Safety Manifolds for Zero-Shot LLM Guardrails

Abstract

Current LLM safety guardrails face a fundamental tension: fine-tuning distorts pre-trained representations while generative judges incur prohibitive inference costs. We challenge the prevailing paradigm by asking: can safety be achieved through pure geometric reasoning over frozen semantic representations? We present HoloAegis, a minimally parametric topological inference framework that decouples representation from reasoning. We term our approach minimally parametric because the only free param

Categories

@misc{li2026holoaegis,
  title = {{HoloAegis: Frozen Representation, Topological Inference: Minimally Parametric Safety Manifolds for Zero-Shot LLM Guardrails}},
  author = {T. Li and Kaijie Liu and Lik-Hang Lee and King-Chung Ho and Ping Shum and Michael K. Ng},
  year = {2026},
  month = aug,
  eprint = {2608.08485},
  archivePrefix = {arXiv},
  url = {https://www.semanticscholar.org/paper/8d4c68176fa54a5bf07893d8029fd1315f37edf9}
}