一作ACL 2026Poster· Natural Language Processing
Unlearners Can Lie: Evaluating and Improving Honesty in LLM Unlearning
Renjie Gu · Jiazhen Du · Yihua Zhang · Sijia Liu
摘要
Unlearning in large language models (LLMs) aims to remove harmful training data while preserving overall utility. However, we find that existing methods often hallucinate, generate abnormal token sequences, or behave inconsistently, raising safety and trust concerns. According to prior literature...