English
Related papers

Related papers: Refusal Falls off a Cliff: How Safety Alignment Fa…

200 papers

Multimodal large language models (MLLMs) have become the cornerstone of today's generative AI ecosystem, sparking intense competition among tech giants and startups. In particular, an MLLM generates a text response given a prompt consisting…

Cryptography and Security · Computer Science 2024-09-09 Zedian Shao , Hongbin Liu , Yuepeng Hu , Neil Zhenqiang Gong

Large reasoning models (e.g., R1, o3) have demonstrated remarkable mathematical problem-solving abilities. However, the high reported accuracy of these advanced models on popular datasets, reliance on purely numerical evaluation and…

Artificial Intelligence · Computer Science 2025-12-10 Dadi Guo , Jiayu Liu , Zhiyuan Fan , Zhitao He , Haoran Li , Yuxin Li , Yumeng Wang , Yi R. Fung

Large Language Models (LLMs) are central to a multitude of applications but struggle with significant risks, notably in generating harmful content and biases. Drawing an analogy to the human psyche's conflict between evolutionary survival…

Computation and Language · Computer Science 2023-11-16 Zi Yin , Wei Ding , Jia Liu

Despite the success of test-time scaling, Large Reasoning Models (LRMs) frequently encounter repetitive loops that lead to computational waste and inference failure. In this paper, we identify a distinct failure mode termed Circular…

Artificial Intelligence · Computer Science 2026-01-12 Zenghao Duan , Liang Pang , Zihao Wei , Wenbin Duan , Yuxin Tian , Shicheng Xu , Jingcheng Deng , Zhiyi Yin , Xueqi Cheng

Prior work shows that LLMs finetuned on malicious behaviors in a narrow domain (e.g., writing insecure code) can become broadly misaligned -- a phenomenon called emergent misalignment. We investigate whether this extends from conventional…

Machine Learning · Computer Science 2025-07-11 James Chua , Jan Betley , Mia Taylor , Owain Evans

Fine-tuning safety-aligned language models for downstream tasks often leads to substantial degradation of refusal behavior, making models vulnerable to adversarial misuse. While prior work has shown that safety-relevant features are encoded…

Machine Learning · Computer Science 2026-05-05 Sadia Asif , Mohammad Mohammadi Amiri

Safety-aligned language models often refuse cybersecurity requests whose wording resembles misuse, even when the task is authorized and defensive. This makes security evaluation ambiguous: a failed answer may reflect missing capability or…

Cryptography and Security · Computer Science 2026-05-19 Isaac David , Arthur Gervais

The rapid development of large language models (LLMs) has not only provided numerous opportunities but also presented significant challenges. This becomes particularly evident when LLMs inadvertently generate harmful or toxic content,…

Computation and Language · Computer Science 2024-02-20 Kai Chen , Chunwei Wang , Kuo Yang , Jianhua Han , Lanqing Hong , Fei Mi , Hang Xu , Zhengying Liu , Wenyong Huang , Zhenguo Li , Dit-Yan Yeung , Lifeng Shang , Xin Jiang , Qun Liu

Large language models (LLMs) have demonstrated remarkable capabilities in complex reasoning tasks. However, they remain highly susceptible to jailbreak attacks that undermine their safety alignment. Existing defense mechanisms typically…

Cryptography and Security · Computer Science 2026-03-17 Yu Pan , Wenlong Yu , Tiejun Wu , Xiaohu Ye , Qiannan Si , Guangquan Xu , Bin Wu

While explicit Chain-of-Thought (CoT) empowers large reasoning models (LRMs), it enables the generation of riskier final answers. Current alignment paradigms primarily rely on externally enforced compliance, optimizing models to detect…

Artificial Intelligence · Computer Science 2026-05-12 Yi Zhang , Yuxin Chen , Leheng Sheng , Dongcheng Zhang , Chaochao Lu , Xiang Wang , An Zhang

The safety alignment of large language models (LLMs) can be circumvented through adversarially crafted inputs, yet the mechanisms by which these attacks bypass safety barriers remain poorly understood. Prior work suggests that a single…

Shojaee et al. (2025) report that Large Reasoning Models (LRMs) exhibit "accuracy collapse" on planning puzzles beyond certain complexity thresholds. We demonstrate that their findings primarily reflect experimental design limitations…

Artificial Intelligence · Computer Science 2025-06-18 A. Lawsen

Large language models (LLMs) aligned for safety often suffer from over-refusal, the tendency to reject seemingly toxic or benign prompts by misclassifying them as toxic. This behavior undermines models' helpfulness and restricts usability…

Computation and Language · Computer Science 2026-03-05 Yuxiao Lu , Lin Xu , Yang Sun , Wenjun Li , Jie Shi

Humans do not just find mistakes after the fact -- we often catch them mid-stream because 'reflection' is tied to the goal and its constraints. Today's large language models produce reasoning tokens and 'reflective' text, but is it…

Artificial Intelligence · Computer Science 2025-10-24 Sion Weatherhead , Flora Salim , Aaron Belbasis

Large Reasoning Models (LRMs) have recently demonstrated impressive performances across diverse domains. However, how the safety of Large Language Models (LLMs) benefits from enhanced reasoning capabilities against jailbreak queries remains…

Computation and Language · Computer Science 2025-09-23 Junda Zhu , Lingyong Yan , Shuaiqiang Wang , Dawei Yin , Lei Sha

Large language models require continuous adaptation to new tasks while preserving safety alignment. However, fine-tuning on even benign data often compromises safety behaviors, including refusal of harmful requests, truthfulness, and…

Machine Learning · Computer Science 2026-04-21 Thong Bach , Dung Nguyen , Thao Minh Le , Truyen Tran

Large Language Models (LLMs) are becoming a prominent generative AI tool, where the user enters a query and the LLM generates an answer. To reduce harm and misuse, efforts have been made to align these LLMs to human values using advanced…

Cryptography and Security · Computer Science 2024-11-08 Xiaomeng Hu , Pin-Yu Chen , Tsung-Yi Ho

Situational awareness, the capacity of an AI system to recognize its own nature, understand its training and deployment context, and reason strategically about its circumstances, is widely considered among the most dangerous emergent…

Artificial Intelligence · Computer Science 2026-03-11 Subramanyam Sahoo , Aman Chadha , Vinija Jain , Divya Chaudhary

Do large language models (LLMs) display rational reasoning? LLMs have been shown to contain human biases due to the data they have been trained on; whether this is reflected in rational reasoning remains less clear. In this paper, we answer…

Computation and Language · Computer Science 2024-02-16 Olivia Macmillan-Scott , Mirco Musolesi

Reinforcement Learning from Human Feedback (RLHF) aligns Large Language Models (LLMs) with human preferences, yet the underlying reward signals they internalize remain hidden, posing a critical challenge for interpretability and safety.…

Machine Learning · Computer Science 2026-01-21 Nyal Patel , Matthieu Bou , Arjun Jagota , Satyapriya Krishna , Sonali Parbhoo