Improving Safety Alignment via Balanced Direct Preference Optimization

Computer Science > Artificial Intelligence [Submitted on 24 Mar 2026] Improving Safety Alignment via Balanced Direct Preference Optimization Shiji Zhao, Mengyang Wang, Shukun Xiong, Fangzhou Chen, Qihui Zhu, Shouwei Ruan, Yisong Xiao, Ranjie Duan, Xun Chen, XingXing Wei With the rapid development and widespread application of Large Language Models (LLMs), their potential safety risks have attracted widespread attention. Reinforcement Learning from Human Feedback (RLHF) has been adopted to enhance the safety performance of LLMs. As a simple and effective alternative to RLHF, Direct Preference Optimization (DPO) is widely used for safety alignment. However, safety alignment still suffers from severe overfitting, which limits its actual performance. This paper revisits the overfitting phenomenon from the perspective of the model's comprehension of the training data. We find that the Imbalanced Preference Comprehension phenomenon exists between responses in preference pairs, which compromises the model's safety performance. To address this, we propose Balanced Direct Preference Optimization (B-DPO), which adaptively modulates optimization strength between preferred and dispreferred responses based on mutual information. A series of experimental results show that B-DPO can enhance the safety capability while maintaining the competitive general capabilities of LLMs on various mainstream benchmarks compared to state-of-the-art methods. \color{red}{Warning: This paper contains examples of harmful texts, and reader discretion is recommended. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2603.22829 [cs.AI]   (or arXiv:2603.22829v1 [cs.AI] for this version)   https://doi.org/10.48550/arXiv.2603.22829 Focus to learn more Submission history From: Mengyang Wang [view email] [v1] Tue, 24 Mar 2026 06:04:09 UTC (2,419 KB) Access Paper: HTML (experimental) view license Current browse context: cs.AI < prev   |   next > new | recent | 2026-03 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)