English
Related papers

Related papers: Steering Externalities: Benign Activation Steering…

200 papers

Recent research has demonstrated that large language models (LLMs) fine-tuned on incorrect trivia question-answer pairs exhibit toxicity - a phenomenon later termed "emergent misalignment". Moreover, research has shown that LLMs possess…

Computation and Language · Computer Science 2026-02-17 Laurène Vaugrante , Anietta Weckauff , Thilo Hagendorff

Large Vision-Language Models (LVLMs) rely on attention-based retrieval of safety instructions to maintain alignment during generation. Existing attacks typically optimize image perturbations to maximize harmful output likelihood, but suffer…

Computer Vision and Pattern Recognition · Computer Science 2026-04-14 Jingru Li , Wei Ren , Tianqing Zhu

Large language models can resist task-misaligned activation steering during inference, sometimes recovering mid-generation to produce improved responses even when steering remains active. We term this Endogenous Steering Resistance (ESR).…

Large language models (LLMs) tend to verbalize confidence scores that are largely detached from their actual accuracy, yet the geometric relationship governing this behavior remain poorly understood. In this work, we present a mechanistic…

Computation and Language · Computer Science 2026-04-02 Miranda Muqing Miao , Lyle Ungar

Activation engineering is becoming increasingly popular as a means of online control of large language models (LLMs). In this work, we extend the idea of inference-time steering with vectors that represent a behavioral direction of interest…

Machine Learning · Computer Science 2024-11-26 Christopher M. Ackerman

Steering vectors (SVs) have emerged as a promising approach for interpreting and controlling LLMs, but current methods typically require large contrastive datasets that are often impractical to construct and may capture spurious…

Machine Learning · Computer Science 2025-08-14 Jacob Dunefsky , Arman Cohan

Large language models remain vulnerable to jailbreak attacks -- inputs designed to bypass safety mechanisms and elicit harmful responses -- despite advances in alignment and instruction tuning. We propose Head-Masked Nullspace Steering…

Cryptography and Security · Computer Science 2026-04-14 Vishal Pramanik , Maisha Maliha , Susmit Jha , Sumit Kumar Jha

Large Language Models (LLMs) exhibit remarkable capabilities across various tasks, yet guiding them to follow desired behaviours during inference remains a significant challenge. Activation steering offers a promising method to control the…

Computation and Language · Computer Science 2025-09-29 Weixuan Wang , Minghao Wu , Barry Haddow , Alexandra Birch

Inference-time LLM alignment methods, particularly activation steering, offer an alternative to fine-tuning by directly modifying activations during generation. Existing methods, however, often rely on non-anticipative interventions that…

Machine Learning · Computer Science 2026-04-22 Julian Skifstad , Xinyue Annie Yang , Glen Chou

Large language models (LLMs) are now deployed at unprecedented scale, assisting millions of users in daily tasks. However, the risk of these models assisting unlawful activities remains underexplored. In this study, we define this high-risk…

Computers and Society · Computer Science 2025-11-27 Xing Wang , Huiyuan Xie , Yiyan Wang , Chaojun Xiao , Huimin Chen , Holli Sargeant , Felix Steffek , Jie Shao , Zhiyuan Liu , Maosong Sun

Large language models (LLMs) increasingly operate in multi-agent and safety-critical settings, raising open questions about how their vulnerabilities scale when models interact adversarially. This study examines whether larger models can…

Machine Learning · Computer Science 2026-01-05 Samuel Nathanson , Rebecca Williams , Cynthia Matuszek

Safety risks of AI models have been widely studied at deployment time, such as jailbreak attacks that elicit harmful outputs. In contrast, safety risks emerging during training remain largely unexplored. Beyond explicit reward hacking that…

Computation and Language · Computer Science 2026-02-05 Zhexin Zhang , Yida Lu , Junfeng Fang , Junxiao Yang , Shiyao Cui , Hao Zhou , Fandong Meng , Jie Zhou , Hongning Wang , Minlie Huang , Tat-Seng Chua

Detecting jailbreak attempts in clinical training large language models (LLMs) requires accurate modeling of linguistic deviations that signal unsafe or off-task user behavior. Prior work on the 2-Sigma clinical simulation platform showed…

Artificial Intelligence · Computer Science 2026-02-17 Tri Nguyen , Huy Hoang Bao Le , Lohith Srikanth Pentapalli , Laurah Turner , Kelly Cohen

Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of linear steering methods, which assume additive behavior in…

Machine Learning · Computer Science 2026-04-07 Narmeen Oozeer , Luke Marks , Shreyans Jain , Fazl Barez , Amirali Abdullah

Linear activation steering is a powerful approach for eliciting the capabilities of large language models and specializing their behavior using limited labeled data. While effective, existing methods often apply a fixed steering strength to…

Computation and Language · Computer Science 2026-04-28 Brandon Hsu , Daniel Beaglehole , Adityanarayanan Radhakrishnan , Mikhail Belkin

Large Language Models (LLMs) face prominent security risks from jailbreaking, a practice that manipulates models to bypass built-in security constraints and generate unethical or unsafe content. Among various jailbreak techniques,…

Cryptography and Security · Computer Science 2026-04-14 Yihao Zhang , Kai Wang , Jiangrong Wu , Haolin Wu , Yuxuan Zhou , Zeming Wei , Dongxian Wu , Xun Chen , Jun Sun , Meng Sun

We show that iterative deployment of large language models (LLMs), each fine-tuned on data carefully curated by users from the previous models' deployment, can significantly change the properties of the resultant models. By testing this…

Artificial Intelligence · Computer Science 2026-01-01 Augusto B. Corrêa , Yoav Gelberg , Luckeciano C. Melo , Ilia Shumailov , André G. Pereira , Yarin Gal

Although value-aligned language models (LMs) appear unbiased in explicit bias evaluations, they often exhibit stereotypes in implicit word association tasks, raising concerns about their fair usage. We investigate the mechanisms behind this…

Computation and Language · Computer Science 2025-06-10 Lihao Sun , Chengzhi Mao , Valentin Hofmann , Xuechunzi Bai

Augmenting Large Language Models (LLMs) with image-understanding capabilities has resulted in a boom of high-performing Vision-Language models (VLMs). While studying the alignment of LLMs to human values has received widespread attention,…

Computer Vision and Pattern Recognition · Computer Science 2024-05-08 Georgios Pantazopoulos , Amit Parekh , Malvina Nikandrou , Alessandro Suglia

Responsible deployment of language models requires mechanisms for refusing unsafe prompts while preserving model performance. While most approaches modify model weights through additional training, we explore an alternative: steering model…

‹ Prev 1 8 9 10 Next ›