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

Weixiang Zhao, Yulin Hu, Zhuojun Li, Yang Deng, Jiahe Guo, Xingyu Sui, Yanyan Zhao, Bing Qin, Tat-Seng Chua, Ting Liu

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 \textsc{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 \textsc{SafePatching} achieves a more comprehensive PSA than baseline methods, further optimizing the balance between being helpful and harmless in current aligned LLMs. Also, \textsc{SafePatching} demonstrates its superiority in continual PSA scenarios.
BibTeX
@misc{zhao2024towards,
  author = {Weixiang Zhao and Yulin Hu and Zhuojun Li and Yang Deng and Jiahe Guo and Xingyu Sui and Yanyan Zhao and Bing Qin and Tat-Seng Chua and Ting Liu},
  title = {{Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching}},
  year = {2024},
  eprint = {2405.13820},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2405.13820},
}

Loading related content...

Figure