paper-with-me

Papers

Silencing the Guardrails: Inference-Time Jailbreaking via Dynamic Contextual Representation Ablation

2026-04-09 · Wenpeng Xing, Moran Fang, Guangtai Wang, Changting Lin, Meng Han arxiv

While Large Language Models (LLMs) have achieved remarkable performance, they remain vulnerable to jailbreak attacks that circumvent safety constraints. Existing strategies, ranging from heuristic prompt engineering to computationally intensive optimization, often face significant trade-offs between effectiveness and efficiency. In this work, we propose Contextual Representation Ablation (CRA), a novel inference-time intervention framework designed to dynamically silence model guardrails. Predicated on the geometric insight that refusal behaviors are mediated by specific low-rank subspaces within the model's hidden states, CRA identifies and suppresses these refusal-inducing activation patterns during decoding without requiring expensive parameter updates or training. Empirical evaluation across multiple safety-aligned open-source LLMs demonstrates that CRA significantly outperforms baselines. These results expose the intrinsic fragility of current alignment mechanisms, revealing that safety constraints can be surgically ablated from internal representations, and underscore the urgent need for more robust defenses that secure the model's latent space.

📄 PDF Abstract BibTeX arXiv:2604.07835

Code (0)

등록된 구현이 없습니다.

Tasks

Prompt Engineering

Similar Papers 제목 키워드 기반

Benchmarking LLM Guardrails in Handling Multilingual Toxicity

2024-10-29 · Yahan Yang, Soham Dan, Dan Roth, Insup Lee

With the ubiquity of Large Language Models (LLMs), guardrails have become crucial to detect and defend against toxic content. However, with the increasing pervasiveness of LLMs in multilingual scenarios, their effectiven…

Benchmarking

Jailbreaking Large Language Models Against Moderation Guardrails via Cipher Characters

2024-05-30 · Haibo Jin, Andy Zhou, Joe D. Menke, Haohan Wang

Large Language Models (LLMs) are typically harmless but remain vulnerable to carefully crafted prompts known as ``jailbreaks'', which can bypass protective measures and induce harmful behavior. Recent advancements in LLM…

Red Teaming

Jailbreaking Safeguarded Text-to-Image Models via Large Language Models

2025-03-03 · Zhengyuan Jiang, Yuepeng Hu, Yuchen Yang, Yinzhi Cao 외

Text-to-Image models may generate harmful content, such as pornographic images, particularly when unsafe prompts are submitted. To address this issue, safety filters are often added on top of text-to-image models, or the…

Language ModelingLanguage ModellingLarge Language Model

Self-Jailbreaking: Language Models Can Reason Themselves Out of Safety Alignment After Benign Reasoning Training

2025-10-23 · Zheng-Xin Yong, Stephen H. Bach arxiv

We discover a novel and surprising phenomenon of unintentional misalignment in reasoning language models (RLMs), which we call self-jailbreaking. Specifically, after benign reasoning training on math or code domains, RLM…

Safety Guardrails for LLM-Enabled Robots

2025-03-10 · Zachary Ravichandran, Alexander Robey, Vijay Kumar, George J. Pappas 외

Although the integration of large language models (LLMs) into robotics has unlocked transformative capabilities, it has also introduced significant safety concerns, ranging from average-case LLM errors (e.g., hallucinati…