English
Related papers

Related papers: Superficial Safety Alignment Hypothesis

200 papers

Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components. However, existing safety attribution methods…

Machine Learning · Computer Science 2026-03-25 Miao Yu , Siyuan Fu , Moayad Aloqaily , Zhenhong Zhou , Safa Otoum , Xing fan , Kun Wang , Yufei Guo , Qingsong Wen

Current safety alignment techniques for large language models (LLMs) face two key challenges: (1) under-generalization, which leaves models vulnerable to novel jailbreak attacks, and (2) over-alignment, which leads to the excessive refusal…

Computation and Language · Computer Science 2025-04-15 Yutao Mou , Yuxiao Luo , Shikun Zhang , Wei Ye

Safety fine-tuning helps align Large Language Models (LLMs) with human preferences for their safe deployment. To better understand the underlying factors that make models safe via safety fine-tuning, we design a synthetic data generation…

Machine Learning · Computer Science 2024-08-22 Samyak Jain , Ekdeep Singh Lubana , Kemal Oksuz , Tom Joy , Philip H. S. Torr , Amartya Sanyal , Puneet K. Dokania

Safety alignment is critical for deploying large language models (LLMs) in real-world applications, yet most existing approaches rely on large human-annotated datasets and static red-teaming benchmarks that are costly, difficult to scale,…

Computation and Language · Computer Science 2026-03-10 Punyajoy Saha , Sudipta Halder , Debjyoti Mondal , Subhadarshi Panda

Large Language Models (LLMs) rely on safety alignment to produce socially acceptable responses. However, this behavior is known to be brittle: further fine-tuning, even on benign or lightly contaminated data, can degrade safety and…

Machine Learning · Computer Science 2026-02-10 Kaustubh Ponkshe , Shaan Shah , Raghav Singhal , Praneeth Vepakomma

Safety alignment is a key requirement for building reliable Artificial General Intelligence. Despite significant advances in safety alignment, we observe that minor latent shifts can still trigger unsafe responses in aligned models. We…

Machine Learning · Computer Science 2025-06-23 Tianle Gu , Kexin Huang , Zongqi Wang , Yixu Wang , Jie Li , Yuanqi Yao , Yang Yao , Yujiu Yang , Yan Teng , Yingchun Wang

Fine-tuning large language models (LLMs) on additional datasets is often necessary to optimize them for specific downstream tasks. However, existing safety alignment measures, which restrict harmful behavior during inference, are…

Computation and Language · Computer Science 2024-10-15 Minjun Zhu , Linyi Yang , Yifan Wei , Ningyu Zhang , Yue Zhang

Safety alignment -- training large language models (LLMs) to refuse harmful requests while remaining helpful -- is critical for responsible deployment. Prior work established that safety behaviors are governed by low-rank structures,…

Computation and Language · Computer Science 2026-01-06 Dianyun Wang , Qingsen Ma , Yuhu Shang , Zhifeng Lu , Zhenbo Xu , Lechen Ning , Huijia Wu , Zhaofeng He

This paper presents a comprehensive empirical study on the safety alignment capabilities. We evaluate what matters for safety alignment in LLMs and LRMs to provide essential insights for developing more secure and reliable AI systems. We…

Computation and Language · Computer Science 2026-02-25 Xing Li , Hui-Ling Zhen , Lihao Yin , Xianzhi Yu , Zhenhua Dong , Mingxuan Yuan

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,…

Computation and Language · Computer Science 2024-12-18 Weixiang Zhao , Yulin Hu , Zhuojun Li , Yang Deng , Jiahe Guo , Xingyu Sui , Yanyan Zhao , Bing Qin , Tat-Seng Chua , Ting Liu

Large Language Models (LLMs) are increasingly used for decision making in embodied agents, yet existing safety evaluations often rely on coarse success rates and domain-specific setups, making it difficult to diagnose why and where these…

Artificial Intelligence · Computer Science 2025-05-27 Yejin Son , Minseo Kim , Sungwoong Kim , Seungju Han , Jian Kim , Dongju Jang , Youngjae Yu , Chanyoung Park

Safety for Large Language Models (LLMs) has been an ongoing research focus since their emergence and is even more relevant nowadays with the increasing capacity of those models. Currently, there are several guardrails in place for all…

Computation and Language · Computer Science 2025-12-25 Eduard Stefan Dinuta , Iustin Sirbu , Traian Rebedea

Safety alignment in Language Models (LMs) is fundamental for trustworthy AI. However, while different stakeholders are trying to leverage Arabic Language Models (ALMs), systematic safety evaluation of ALMs remains largely underexplored,…

Computation and Language · Computer Science 2026-03-06 Omar Abdelnasser , Fatemah Alharbi , Khaled Khasawneh , Ihsen Alouani , Mohammed E. Fouda

Fine-tuning safety-aligned large language models (LLMs) can substantially compromise their safety. Previous approaches require many safety samples or calibration sets, which not only incur significant computational overhead during…

Machine Learning · Computer Science 2026-01-07 Jiawen Zhang , Lipeng He , Kejia Chen , Jian Lou , Jian Liu , Xiaohu Yang , Ruoxi Jia

Despite the remarkable proficiency of \textit{Large Reasoning Models} (LRMs) in handling complex reasoning tasks, their reliability in safety-critical scenarios remains uncertain. Existing evaluations primarily assess response-level safety,…

Artificial Intelligence · Computer Science 2025-05-27 Baihui Zheng , Boren Zheng , Kerui Cao , Yingshui Tan , Zhendong Liu , Weixun Wang , Jiaheng Liu , Jian Yang , Wenbo Su , Xiaoyong Zhu , Bo Zheng , Kaifu Zhang

We identify a structural weakness in current large language model (LLM) alignment: modern refusal mechanisms are fail-open. While existing approaches encode refusal behaviors across multiple latent features, suppressing a single dominant…

Machine Learning · Computer Science 2026-02-20 Zachary Coalson , Beth Sohler , Aiden Gabriel , Sanghyun Hong

The safety of large language models (LLMs) has increasingly emerged as a fundamental aspect of their development. Existing safety alignment for LLMs is predominantly achieved through post-training methods, which are computationally…

Artificial Intelligence · Computer Science 2026-02-03 Sicheng Shen , Mingyang Lv , Han Shen , Jialin Wu , Binghao Wang , Zhou Yang , Guobin Shen , Dongcheng Zhao , Feifei Zhao , Yi Zeng

Large language models (LLMs) exhibit advancing capabilities in complex tasks, such as reasoning and graduate-level question answering, yet their resilience against misuse, particularly involving scientifically sophisticated risks, remains…

Small language models (SLMs) are increasingly deployed on edge devices, making their safety alignment crucial yet challenging. Current shallow alignment methods that rely on direct refusal of malicious queries fail to provide robust…

Cryptography and Security · Computer Science 2025-11-11 Haonan Shi , Guoli Wang , Tu Ouyang , An Wang

Large language models (LLMs) and multimodal LLMs are typically safety-aligned before release to prevent harmful content generation. However, recent studies show that safety behaviors are concentrated in a small subset of parameters, making…

Machine Learning · Computer Science 2026-02-13 Zhaoxin Wang , Jiaming Liang , Fengbin Zhu , Weixiang Zhao , Junfeng Fang , Jiayi Ji , Handing Wang , Tat-Seng Chua