English
Related papers

Related papers: Refusal in Language Models Is Mediated by a Single…

200 papers

Multimodal large language models (MLLMs) often fail to transfer safety capabilities learned in the text modality to semantically equivalent non-text inputs, revealing a persistent multimodal safety gap. We study this gap from a…

Artificial Intelligence · Computer Science 2026-05-19 Jiahe Guo , Xiangran Guo , Jiaxuan Chen , Weixiang Zhao , Yanyan Zhao , Yutai Hou , Qianchao Wang , Dandan Tu , Bing Qin

We find that language models have difficulties generating fallacious and deceptive reasoning. When asked to generate deceptive outputs, language models tend to leak honest counterparts but believe them to be false. Exploiting this…

Computation and Language · Computer Science 2025-05-26 Yue Zhou , Henry Peng Zou , Barbara Di Eugenio , Yang Zhang

Large language models (LLMs) achieve impressive performance across diverse tasks yet remain vulnerable to jailbreak attacks that bypass safety mechanisms. We present RAID (Refusal-Aware and Integrated Decoding), a framework that…

Computation and Language · Computer Science 2025-12-23 Tuan T. Nguyen , John Le , Thai T. Vu , Willy Susilo , Heath Cooper

When a user tells an AI system that someone "should not" take an action, the system ought to treat this as a prohibition. Yet many large language models do the opposite: they interpret negated instructions as affirmations. We audited 16…

Artificial Intelligence · Computer Science 2026-01-30 Katherine Elkins , Jon Chun

Large Language Models(LLMs) have been successful in numerous fields. Alignment has usually been applied to prevent them from harmful purposes. However, aligned LLMs remain vulnerable to jailbreak attacks that deliberately mislead them into…

Cryptography and Security · Computer Science 2026-02-17 Shang Liu , Hanyu Pei , Zeyan Liu

Multimodal Large Language Models (MLLMs) have showcased impressive performance in a variety of multimodal tasks. On the other hand, the integration of additional image modality may allow the malicious users to inject harmful content inside…

Cryptography and Security · Computer Science 2025-04-23 Yulin Chen , Haoran Li , Yirui Zhang , Zihao Zheng , Yangqiu Song , Bryan Hooi

Inasmuch as the removal of refusal behavior from instruction-tuned language models by directional abliteration requires the extraction of refusal-mediating directions from the residual stream activation space, and inasmuch as the…

Machine Learning · Computer Science 2026-03-24 Valentin Petrov

We propose affine concept editing (ACE) as an approach for steering language models' behavior by intervening directly in activations. We begin with an affine decomposition of model activation vectors and show that prior methods for steering…

Machine Learning · Computer Science 2025-01-29 Thomas Marshall , Adam Scherlis , Nora Belrose

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

In this paper, we present empirical studies on conversational recommendation tasks using representative large language models in a zero-shot setting with three primary contributions. (1) Data: To gain insights into model behavior in…

Information Retrieval · Computer Science 2023-08-22 Zhankui He , Zhouhang Xie , Rahul Jha , Harald Steck , Dawen Liang , Yesu Feng , Bodhisattwa Prasad Majumder , Nathan Kallus , Julian McAuley

Multi-turn jailbreaks exploit the ability of large language models to accumulate and act on conversational context. Instead of stating a harmful request directly, an attacker can gradually steer the conversation toward an unsafe answer.…

Cryptography and Security · Computer Science 2026-05-13 Xinkai Zhang , Zhipeng Wei , Huanli Gong , Jing Ting Zheng , Yuchen Zhang , Yue Dong , N. Benjamin Erichson

Multimodal AI systems integrate text generation, image generation, and other capabilities within a single conversational interface. These systems employ safety mechanisms to prevent disallowed actions, including the removal of watermarks…

Computers and Society · Computer Science 2026-01-13 Bentley DeVilling

Representation Engineering analyses often characterize refusal using static directions extracted from terminal or pooled representations. We ask whether this view misses how refusal is constructed across layer-token positions. Using causal…

Cryptography and Security · Computer Science 2026-05-27 Xulin Hu , Che Wang , Wei Yang Bryan Lim , Jianbo Gao , Zhong Chen

The deployment of Large Language Models (LLMs) as tool-using agents causes their alignment training to manifest in new ways. Recent work finds that language models can use tools in ways that contradict the interests or explicit instructions…

Machine Learning · Computer Science 2026-04-24 Kushal Agrawal , Frank Xiao , Guido Bergman , Asa Cooper Stickland

We present MultiBreak, a scalable and diverse multi-turn jailbreak benchmark to evaluate large language model (LLM) safety. Multi-turn jailbreaks mimic natural conversational settings, making them easier to bypass safety-aligned LLM than…

Computation and Language · Computer Science 2026-05-05 Jialin Song , Xiaodong Liu , Weiwei Yang , Wuyang Chen , Mingqian Feng , Xuekai Zhu , Jianfeng Gao

Large Language Models (LLMs) exhibit strong conversational abilities but often generate falsehoods. Prior work suggests that the truthfulness of simple propositions can be represented as a single linear direction in a model's internal…

Machine Learning · Computer Science 2025-05-29 Stanley Yu , Vaidehi Bulusu , Oscar Yasunaga , Clayton Lau , Cole Blondin , Sean O'Brien , Kevin Zhu , Vasu Sharma

Current LLMs are trained to refuse potentially harmful input queries regardless of whether users actually had harmful intents, causing a tradeoff between safety and user experience. Through a study of 480 participants evaluating 3,840…

Computation and Language · Computer Science 2025-12-04 Mingqian Zheng , Wenjia Hu , Patrick Zhao , Motahhare Eslami , Jena D. Hwang , Faeze Brahman , Carolyn Rose , Maarten Sap

Pretrained large language models (LLMs) are becoming increasingly powerful and ubiquitous in mainstream applications such as being a personal assistant, a dialogue model, etc. As these models become proficient in deducing user preferences…

Computation and Language · Computer Science 2023-02-22 Varshini Subhash

Language models can be persuaded to abandon factual knowledge. This vulnerability is central to AI safety, but its internal mechanism remains poorly understood. We uncover a compact causal mechanism for persuasion-induced factual errors. A…

Artificial Intelligence · Computer Science 2026-05-12 Xiangkun Sun , Lingkai Kong , Aoqi Zhang , Liang Zeng , Tonghan Wang

The increased presence of large language models (LLMs) in educational settings has ignited debates concerning negative repercussions, including overreliance and inadequate task reflection. Our work advocates moderated usage of such models,…

‹ Prev 1 8 9 10 Next ›