Exploring and Developing a Pre-Model Safeguard with Draft Models

Abstract

Large Language Model (LLM) alignment remains vulnerable to jailbreak attacks that elicit unsafe responses, motivating pre-model and post-model guards. Pre-model guards audit the safety of prompts before invoking target models. However, relying solely on the prompt often leads to high false-negative rates (i.e., jailbreak attacks go undetected). Post-model guards address this issue by auditing both the user prompt and the target model's response. However, they incur a high computational cost, inc

Categories

Framework mappings

Suggested from the entry's categories.

@misc{cai2026exploring,
  title = {{Exploring and Developing a Pre-Model Safeguard with Draft Models}},
  author = {Hongyu Cai and Arjun Arunasalam and Yiming Liang and Antonio Bianchi and Z. Berkay Celik},
  year = {2026},
  month = may,
  eprint = {2605.19321},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.19321}
}