---
title: "Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching"
authors:
  - "Weixiang Zhao"
  - "Yulin Hu"
  - "Yang Deng"
  - "Jiahe Guo"
  - "Xingyu Sui"
  - "Yanyan Zhao"
  - "Bing Qin"
  - "Ting Liu"
doi: "10.24963/ijcai.2026/99"
url: "https://ijcai.org/proceedings/2026/99"
pdf: "https://ijcai.org/proceedings/2026/0099.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "880-888"
track: "Main Track"
keywords:
  - "AI Ethics, Trust, Fairnes: Safety and robustness"
  - "Natural Language Processing: Language models"
---

# Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching

Weixiang Zhao, Yulin Hu, Yang Deng, Jiahe Guo, Xingyu Sui, Yanyan Zhao, Bing Qin, Ting Liu

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 880-888. DOI: [10.24963/ijcai.2026/99](https://doi.org/10.24963/ijcai.2026/99)

## Abstract

Safety alignment of large language models (LLMs) has been gaining increasing attention. However, current safety-aligned LLMs suffer from the fragile and imbalanced safety mechanisms, which can still be induced to generate unsafe responses, exhibit over-safety by rejecting safe user inputs, and fail to preserve general utility after safety alignment. To this end, we propose a novel post safety alignment (PSA) method to address these inherent and emerging safety challenges, including safety enhancement, over-safety mitigation, and utility preservation. In specific, we introduce SafePatching, a novel framework for comprehensive PSA, where two distinct safety patches are developed on the harmful data to enhance safety and mitigate over-safety concerns, and then seamlessly integrated into the target LLM backbone without compromising its utility.  Extensive experiments on four representative aligned LLMs, including LLaMA-2/3, Gemma and Mistral, show that SafePatching achieves a more comprehensive PSA than baseline methods, further optimizing the balance between being helpful and harmless in current aligned LLMs. Also, SafePatching demonstrates its superiority in continual PSA scenarios.