ACL Findings2024
Towards Safer Large Language Models through Machine Unlearning
Zheyuan Liu, Guangyao Dou, Zhaoxuan Tan, Yijun Tian, Meng Jiang
Abstract
The rapid advancement of Large Language Models (LLMs) has demonstrated their vast potential across various domains, attributed to their extensive pretraining knowledge and exceptional generalizability. However, LLMs often encounter challenges in generating harmful content when faced with problematic prompts. To address this problem, existing work attempted to implement a gradient ascent based approach to prevent LLMs from producing harmful output. While these methods can be effective, they frequently impact the model utility in responding to normal prompts. To address this gap, we introduce Selective Knowledge negation Unlearning (SKU), a novel unlearning framework for LLMs, designed to eliminate harmful knowledge while preserving utility on normal prompts. Specifically, SKU is consisted of two stages: harmful knowledge acquisition stage and knowledge negation stage. The first stage aims to identify and acquire harmful knowledge within the model, whereas the second is dedicated to remove this knowledge. SKU selectively isolates and removes harmful knowledge in model parameters, ensuring the model's performance remains robust on normal prompts. Our experiments conducted across various LLM architectures demonstrate that SKU identifies a good balance point between removing harmful information and preserving utility. 1 2
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
46 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Large Language Model Unlearning via Embedding-Corrupted Prompts cites Language Models are Few-Shot Learners
- TruthfulQA: Measuring How Models Mimic Human Falsehoods cites Training language models to follow instructions with human feedback
- Training language models to follow instructions with human feedback cites TruthfulQA: Measuring How Models Mimic Human Falsehoods
- Large Language Models are Zero-Shot Reasoners cites Language Models are Few-Shot Learners
- Large Language Models are Zero-Shot Reasoners cites Training language models to follow instructions with human feedback
- Solving Quantitative Reasoning Problems with Language Models cites Language Models are Few-Shot Learners
- Training language models to follow instructions with human feedback cites Language Models are Few-Shot Learners
- On Effects of Steering Latent Representation for Large Language Model Unlearning cites Large Language Model Unlearning via Embedding-Corrupted Prompts
Cited by top-tier papers46
- Large Language Model Unlearning via Embedding-Corrupted PromptsChris Yuhao Liu, Yaxuan Wang, Jeffrey Flanigan, Yang LiuNeurIPS 2024 · 138 citations
- On Effects of Steering Latent Representation for Large Language Model UnlearningHuu-Tien Dang, Tin Pham, Hoang Thanh-Tung, Naoya InoueAAAI 2025 · 33 citations
- WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language ModelsJinghan Jia, Jiancheng Liu, Yihua Zhang, Parikshit Ram et al.NeurIPS 2024 · 32 citations
- ReLearn: Unlearning via Learning for Large Language ModelsHaoming Xu, Ningyuan Zhao, Liming Yang, Sendong Zhao et al.ACL 2025 · 18 citations
- NGQA: A Nutritional Graph Question Answering Benchmark for Personalized Health-aware Nutritional ReasoningZheyuan Zhang, Yiyang Li, Nhi Ha Lan Le, Zehong Wang et al.ACL 2025 · 18 citations
Builds on21
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Training language models to follow instructions with human feedbackLong Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida et al.NeurIPS 2022 · 24,707 citations
- Large Language Models are Zero-Shot ReasonersTakeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo et al.NeurIPS 2022 · 8,168 citations
- TruthfulQA: Measuring How Models Mimic Human FalsehoodsStephanie Lin, Jacob Hilton, Owain EvansACL 2022 · 3,228 citations
- Solving Quantitative Reasoning Problems with Language ModelsAitor Lewkowycz, Anders Andreassen, David Dohan, Ethan Dyer et al.NeurIPS 2022 · 2,039 citations
Related papers
- Adaptive Localization of Knowledge Negation for Continual LLM UnlearningAbudukelimu Wuerkaixi, Qizhou Wang, Sen Cui, Wutong Xu et al.ICML 2025
- LLM-Eraser: Optimizing Large Language Model Unlearning through Selective PruningShengming Zhang, Le Zhang, Jingbo Zhou, Zhi Zheng et al.KDD 2025 · 2 citations
- Not Every Token Needs Forgetting: Selective Unlearning Balancing Forgetting and Utility in Large Language ModelsYixin Wan, Anil Ramakrishna, Kai-Wei Chang, Volkan Cevher et al.EMNLP Findings 2025
- Safety Alignment via Constrained Knowledge UnlearningZesheng Shi, Yucheng Zhou, Jing Li, Yuxin Jin et al.ACL 2025 · 8 citations
- Model Unlearning via Sparse Autoencoder Subspace Guided ProjectionsXu Wang, Zihao Li, Benyou Wang, Yan Hu et al.EMNLP 2025 · 9 citations
