A Lightweight Explainable Guardrail for Prompt Safety

Md Asiful Islam, Mihai SurdeanuSubmission to ICLR 2026 · Withdrawn

Its authors withdrew this paper, so there is no full text to read or summarize.

Abstract

We propose a lightweight explainable guardrail (LEG) method for the classification of unsafe prompts. LEG uses a multi-task learning architecture to jointly learn a prompt classifier and an explanation classifier, where the latter labels prompt words that explain the safe/unsafe overall decision. LEG is trained using synthetic data for explainability, which is generated using a novel strategy that counteracts the confirmation biases of LLMs. Lastly, LEG's training process uses a novel loss that captures global explanation priors and combines cross-entropy and focal losses with uncertainty-based weighting. LEG obtains equivalent or better performance than the state-of-the-art for both prompt classification and explainability, both in-domain and out-of-domain on three datasets, despite the fact that its model size is considerably smaller than current approaches. If accepted, we will release all models and the annotated dataset publicly.
BibTeX
@misc{islam2026lightweight,
  author = {Md Asiful Islam and Mihai Surdeanu},
  title = {{A Lightweight Explainable Guardrail for Prompt Safety}},
  year = {2026},
  eprint = {2602.15853},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2602.15853},
}
Figure