Faithful Dual-constrained Erasure for Robust LLM Safety Alignment
Machine unlearning has emerged as a crucial mechanism for removing hazardous knowledge and enforcing safety alignment in Large Language Models (LLMs). However, recent studies reveal a persistent security risk: unlearned models remain highly vulnerable to retraining attacks, where suppressed malicious behaviors rapidly...
Jia-Qing Li, Shi-De Zhou, Zhi-Bo Zhang et al.
· 0 citations