English
Related papers

Related papers: Backtracking Improves Generation Safety

200 papers

Backdoor attacks are a significant threat to large language models (LLMs), often embedded via public checkpoints, yet existing defenses rely on impractical assumptions about trigger settings. To address this challenge, we propose…

Computation and Language · Computer Science 2026-05-14 Liang Lin , Miao Yu , Moayad Aloqaily , Zhenhong Zhou , Kun Wang , Linsey Pang , Prakhar Mehrotra , Qingsong Wen

Large Language Models (LLMs) are known to be vulnerable to backdoor attacks, where triggers embedded in poisoned samples can maliciously alter LLMs' behaviors. In this paper, we move beyond attacking LLMs and instead examine backdoor…

Cryptography and Security · Computer Science 2025-02-18 Huaizhi Ge , Yiming Li , Qifan Wang , Yongfeng Zhang , Ruixiang Tang

Safety is one of the main challenges in applying reinforcement learning to realistic environmental tasks. To ensure safety during and after training process, existing methods tend to adopt overly conservative policy to avoid unsafe…

Machine Learning · Computer Science 2023-06-27 Xiao Zhang , Hai Zhang , Hongtu Zhou , Chang Huang , Di Zhang , Chen Ye , Junqiao Zhao

Large language models (LLMs) are susceptible to social-engineered attacks that are human-interpretable but require a high level of comprehension for LLMs to counteract. Existing defensive measures can only mitigate less than half of these…

Computation and Language · Computer Science 2025-05-01 Canaan Yung , Hadi Mohaghegh Dolatabadi , Sarah Erfani , Christopher Leckie

Large Language Models (LLMs) have shown remarkable effectiveness in various general-domain natural language processing (NLP) tasks. However, their performance in transportation safety domain tasks has been suboptimal, primarily attributed…

Computation and Language · Computer Science 2023-07-31 Ou Zheng , Mohamed Abdel-Aty , Dongdong Wang , Chenzhu Wang , Shengxuan Ding

Recent advancements in Large Language Models (LLMs) have significantly enhanced their ability to process long contexts, yet a notable gap remains in generating long, aligned outputs. This limitation stems from a training gap where…

Computation and Language · Computer Science 2024-11-01 Shanghaoran Quan , Tianyi Tang , Bowen Yu , An Yang , Dayiheng Liu , Bofei Gao , Jianhong Tu , Yichang Zhang , Jingren Zhou , Junyang Lin

Despite the intrinsic risk-awareness of Large Language Models (LLMs), current defenses often result in shallow safety alignment, rendering models vulnerable to disguised attacks (e.g., prefilling) while degrading utility. To bridge this…

Cryptography and Security · Computer Science 2026-01-26 Xianya Fang , Xianying Luo , Yadong Wang , Xiang Chen , Yu Tian , Zequn Sun , Rui Liu , Jun Fang , Naiqiang Tan , Yuanning Cui , Sheng-Jun Huang

Large Language Models (LLMs) generating unsafe responses to toxic prompts is a significant issue in their applications. While various efforts aim to address this safety concern, previous approaches often demand substantial human data…

Computation and Language · Computer Science 2024-12-12 Yuxiao Lu , Arunesh Sinha , Pradeep Varakantham

Safety alignment in large language models (LLMs) is achieved through fine-tuning mechanisms that regulate neuron activations to suppress harmful content. In this work, we propose a novel approach to induce disalignment by identifying and…

Machine Learning · Computer Science 2025-05-01 Yi Zhou , Wenpeng Xing , Dezhang Kong , Changting Lin , Meng Han

Larger language models (LLMs) have taken the world by storm with their massive multi-tasking capabilities simply by optimizing over a next-word prediction objective. With the emergence of their properties and encoded knowledge, the risk of…

Computation and Language · Computer Science 2023-08-31 Rishabh Bhardwaj , Soujanya Poria

As the development of large language models (LLMs) rapidly advances, securing these models effectively without compromising their utility has become a pivotal area of research. However, current defense strategies against jailbreak attacks…

Cryptography and Security · Computer Science 2024-12-25 Caishuang Huang , Wanxu Zhao , Rui Zheng , Huijie Lv , Wenyu Zhan , Shihan Dou , Sixian Li , Xiao Wang , Enyu Zhou , Junjie Ye , Yuming Yang , Tao Gui , Qi Zhang , Xuanjing Huang

With adversarial or otherwise normal prompts, existing large language models (LLM) can be pushed to generate toxic discourses. One way to reduce the risk of LLMs generating undesired discourses is to alter the training of the LLM. This can…

Computation and Language · Computer Science 2023-02-28 Meng Cao , Mehdi Fatemi , Jackie Chi Kit Cheung , Samira Shabanian

Large Reasoning Models (LRMs) achieve remarkable success through explicit thinking steps, yet the thinking steps introduce a novel risk by potentially amplifying unsafe behaviors. Despite this vulnerability, conventional defense mechanisms…

Artificial Intelligence · Computer Science 2026-01-08 Su-Hyeon Kim , Hyundong Jin , Yejin Lee , Yo-Sub Han

Likelihood training and maximization-based decoding result in dull and repetitive generated texts even when using powerful language models (Holtzman et al., 2019). Adding a loss function for regularization was shown to improve text…

Computation and Language · Computer Science 2021-01-13 Evgeny Lagutin , Daniil Gavrilov , Pavel Kalaidin

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

While the wide adoption of refusal training in large language models (LLMs) has showcased improvements in model safety, recent works have highlighted shortcomings due to the shallow nature of these alignment methods. To this end, the work…

Machine Learning · Computer Science 2026-04-17 Pankayaraj Pathmanathan , Furong Huang

Text-to-image (T2I) generative models have achieved remarkable visual fidelity, yet remain vulnerable to generating unsafe content. Existing safety defenses typically intervene internally within the generative model, but suffer from severe…

Computer Vision and Pattern Recognition · Computer Science 2026-01-21 Xiangtao Meng , Yingkai Dong , Ning Yu , Li Wang , Zheng Li , Shanqing Guo

We have uncovered a powerful jailbreak technique that leverages large language models' ability to diverge from prior context, enabling them to bypass safety constraints and generate harmful outputs. By simply instructing the LLM to deviate…

Computation and Language · Computer Science 2025-05-13 Weiliang Zhao , Daniel Ben-Levi , Wei Hao , Junfeng Yang , Chengzhi Mao

Autoformalization, the process of transforming informal mathematical language into formal specifications and proofs remains a difficult task for state-of-the-art (large) language models. Existing works point to competing explanations for…

Artificial Intelligence · Computer Science 2025-02-25 Willy Chan , Michael Souliman , Jakob Nordhagen , Brando Miranda , Elyas Obbad , Kai Fronsdal Sanmi Koyejo

Self-supervised pre-training, such as BERT, MASS and BART, has emerged as a powerful technique for natural language understanding and generation. Existing pre-training techniques employ autoencoding and/or autoregressive objectives to train…

Computation and Language · Computer Science 2020-09-22 Bin Bi , Chenliang Li , Chen Wu , Ming Yan , Wei Wang , Songfang Huang , Fei Huang , Luo Si