@inproceedings{zhang-etal-2026-enhance,
    title = "How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study",
    author = "Zhang, Zhexin  and
      Loye, Xian Qi  and
      Huang, Victor Shea-Jay  and
      Yang, Junxiao  and
      Zhu, Qi  and
      Cui, Shiyao  and
      Mi, Fei  and
      Shang, Lifeng  and
      Wang, Yingkang  and
      Wang, Hongning  and
      Huang, Minlie",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.936/",
    pages = "20438--20457",
    ISBN = "979-8-89176-390-6",
    abstract = "Large Reasoning Models (LRMs) have achieved remarkable success on reasoning-intensive tasks such as mathematics and programming. However, their enhanced reasoning capabilities do not necessarily translate to improved safety performance{---}and in some cases, may even degrade it. This raises an important research question: how should we enhance the safety of LRMs? In this paper, we present a comprehensive empirical study on how to enhance the safety of LRMs through Supervised Fine-Tuning (SFT). Our investigation begins with an unexpected observation: directly distilling safe responses from DeepSeek-R1 fails to significantly enhance safety. We analyze this phenomenon and identify five key risky patterns that contribute to it. We then demonstrate that explicitly addressing these issues during the data distillation process can lead to substantial safety improvements. Next, we explore whether a long and complex reasoning process is necessary for achieving safety. Interestingly, we find that simply using short or template-based reasoning process can attain comparable safety performance. These findings prompt a deeper reflection on the role of reasoning in ensuring safety. Finally, we conduct a comprehensive ablation study to reveal the impact of different training configurations. Overall, we hope our empirical study could provide a more holistic picture on enhancing the safety of LRMs."
}