SlotGCG: Exploiting the Positional Vulnerability in LLMs for Jailbreak Attacks

Abstract

As large language models (LLMs) are widely deployed, identifying their vulnerability through jailbreak attacks becomes increasingly critical. Optimization-based attacks like Greedy Coordinate Gradient (GCG) have focused on inserting adversarial tokens to the end of prompts. However, GCG restricts adversarial tokens to a fixed insertion point (typically the prompt suffix), leaving the effect of inserting tokens at other positions unexplored. In this paper, we empirically investigate \emph{slots},

Categories

Framework mappings

Suggested from the entry's categories.

@misc{jeong2026slotgcg,
  title = {{SlotGCG: Exploiting the Positional Vulnerability in LLMs for Jailbreak Attacks}},
  author = {Seungwon Jeong and Jiwoo Jeong and Hyeonjin Kim and Yunseok Lee and Woojin Lee},
  year = {2026},
  month = jun,
  eprint = {2606.05609},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.05609}
}