paperSeptember 2026Unreviewed

Beyond Cross-Lingual Transfer: Benchmarking Propagation Boundaries in Multilingual LLM Unlearning

Pengyang Shao, Chuanpeng Lu, Wei Qin, Yanzheng Jin, Xiaohao Liu, Xi Ai, Kenji Kawaguchi, Richang Hong

Abstract

Large Language Model (LLM) unlearning aims to suppress target knowledge while preserving general capabilities. In multilingual settings, unlearning must additionally propagate within its intended linguistic scope. However, existing evaluations mainly measure cross-lingual transfer and cannot distinguish insufficient from excessive propagation. We introduce CLLPU (Cross-Lingual and Language-Bound Protocol for LLM Unlearning), a multilingual benchmark that formulates this problem through two setti

Cite

@misc{shao2026beyond,
  title = {{Beyond Cross-Lingual Transfer: Benchmarking Propagation Boundaries in Multilingual LLM Unlearning}},
  author = {Pengyang Shao and Chuanpeng Lu and Wei Qin and Yanzheng Jin and Xiaohao Liu and {Xi Ai} and Kenji Kawaguchi and Richang Hong},
  year = {2026},
  month = sep,
  eprint = {2609.05976},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.05976}
}