← Back to all news
arXiv AI By Hamid Dashtbani, Alexandra Dmitrienko

Addition by Subtraction: Toward Mitigating Fragile Safety Alignment in Large Language Models

Read the original on arXiv AI →

The paper introduces a method called Addition by Subtraction to strengthen safety alignment in large language models. By gradually removing identified safety‑critical neurons during finetuning and re‑optimizing preferences, the authors reduce the model’s reliance on sparse components that can be targeted by pruning or ablation attacks. Experiments on four models from three vendors show that this approach improves robustness to both white‑box and black‑box jailbreaks, lowers attack success rates by up to 90%, and enhances truthfulness while preserving general capabilities.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv AI
Aug 26

NeuronGuard: Robust LLM Safety Alignment via Ablation-Aware Safety Signal Redistribution

NeuronGuard is a fine‑tuning defense for large language models that hardens them against both jailbreak and neuron‑level attacks. It redistributes safety signals across many neurons by identifying safety‑critical ones with per‑layer linear classifiers, enforcing refusal behavior when those neurons are ablated, and applying KL‑divergence regularization for consistency. A randomized gradient projection preserves task performance, and the authors provide a formal guarantee that NeuronGuard lowers the attack success rate upper bound, with experiments showing near‑zero success rates across multiple models and attack strategies.

By Anjun Gao, Yueyang Quan, Yufei Xia, Zhuqing Liu, Minghong Fang
llmsfine-tuningmultimodalbenchmarkssafety
More like this →
arXiv AI
Oct 19

Safety Alignment via Constrained Knowledge Unlearning

The paper introduces Constrained Knowledge Unlearning (CKU), a safety alignment method that selectively prunes gradients of neurons in specific MLP layers to delete harmful knowledge while preserving useful information. CKU identifies a subset of neurons associated with valuable knowledge and protects them during the unlearning process, thereby mitigating jailbreak attacks. Experiments show that CKU improves model safety without sacrificing overall performance, and the study also offers insights into neuron knowledge sensitivity across MLP layers.

By Zesheng Shi, Yucheng Zhou, Jing Li
llmssafety
More like this →