Image de couverture de livre

Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Leong, Chak Tou, Yin, Qingyu, Wang, Jian, Li, Wenjie
Format:	Preprint
Publié:	2025
Sujets:	Computation and Language Artificial Intelligence Cryptography and Security
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

No Two Devils Alike: Unveiling Distinct Mechanisms of Fine-tuning Attacks
par: Leong, Chak Tou, et autres
Publié: (2024)

Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?
par: Yin, Qingyu, et autres
Publié: (2025)

AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models
par: Zhang, Jinchuan, et autres
Publié: (2025)

Imitate Before Detect: Aligning Machine Stylistic Preference for Machine-Revised Text Detection
par: Chen, Jiaqi, et autres
Publié: (2024)

CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning
par: Yi, Biao, et autres
Publié: (2025)

Safeguarding Large Language Models in Real-time with Tunable Safety-Performance Trade-offs
par: Fonseca, Joao, et autres
Publié: (2025)

Privacy-Preserving Instructions for Aligning Large Language Models
par: Yu, Da, et autres
Publié: (2024)

GSPR: Aligning LLM Safeguards as Generalizable Safety Policy Reasoners
par: Li, Haoran, et autres
Publié: (2025)

RED QUEEN: Safeguarding Large Language Models against Concealed Multi-Turn Jailbreaking
par: Jiang, Yifan, et autres
Publié: (2024)

S-Eval: Towards Automated and Comprehensive Safety Evaluation for Large Language Models
par: Yuan, Xiaohan, et autres
Publié: (2024)

BitBypass: A New Direction in Jailbreaking Aligned Large Language Models with Bitstream Camouflage
par: Nakka, Kalyan, et autres
Publié: (2025)

Why Are My Prompts Leaked? Unraveling Prompt Extraction Threats in Customized Large Language Models
par: Liang, Zi, et autres
Publié: (2024)

Understanding and Mitigating Over-refusal for Large Language Models via Safety Representation
par: Zhang, Junbo, et autres
Publié: (2025)

EnchTable: Unified Safety Alignment Transfer in Fine-tuned Large Language Models
par: Wu, Jialin, et autres
Publié: (2025)

ML-Bench&Guard: Policy-Grounded Multilingual Safety Benchmark and Guardrail for Large Language Models
par: Zhao, Yunhan, et autres
Publié: (2026)

Representation Bending for Large Language Model Safety
par: Yousefpour, Ashkan, et autres
Publié: (2025)

Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models
par: Shen, Guobin, et autres
Publié: (2024)

One Trigger Token Is Enough: A Defense Strategy for Balancing Safety and Usability in Large Language Models
par: Gu, Haoran, et autres
Publié: (2025)

AgentBnB: A Browser-Based Cybersecurity Tabletop Exercise with Large Language Model Support and Retrieval-Aligned Scaffolding
par: Anwar, Arman, et autres
Publié: (2025)

GLiGuard: Schema-Conditioned Classification for LLM Safeguard
par: Zaratiana, Urchade, et autres
Publié: (2026)

Advancing Adversarial Suffix Transfer Learning on Aligned Large Language Models
par: Liu, Hongfu, et autres
Publié: (2024)

ForgeDAN: An Evolutionary Framework for Jailbreaking Aligned Large Language Models
par: Cheng, Siyang, et autres
Publié: (2025)

CCJA: Context-Coherent Jailbreak Attack for Aligned Large Language Models
par: Zhou, Guanghao, et autres
Publié: (2025)

Rethinking Backdoor Detection Evaluation for Language Models
par: Yan, Jun, et autres
Publié: (2024)

Internal Safety Collapse in Frontier Large Language Models
par: Wu, Yutao, et autres
Publié: (2026)

Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models?
par: Karkevandi, Mohammad Bahrami, et autres
Publié: (2024)

Safeguarding Multimodal Knowledge Copyright in the RAG-as-a-Service Environment
par: Chen, Tianyu, et autres
Publié: (2025)

SGuard-v1: Safety Guardrail for Large Language Models
par: Lee, JoonHo, et autres
Publié: (2025)

Jailbreaking Safeguarded Text-to-Image Models via Large Language Models
par: Jiang, Zhengyuan, et autres
Publié: (2025)

BadApex: Backdoor Attack Based on Adaptive Optimization Mechanism of Black-box Large Language Models
par: Wu, Zhengxian, et autres
Publié: (2025)

QueryAttack: Jailbreaking Aligned Large Language Models Using Structured Non-natural Query Language
par: Zou, Qingsong, et autres
Publié: (2025)

Imposter.AI: Adversarial Attacks with Hidden Intentions towards Aligned Large Language Models
par: Liu, Xiao, et autres
Publié: (2024)

Sanitize Your Responses: Mitigating Privacy Leakage in Large Language Models
par: Fu, Wenjie, et autres
Publié: (2025)

SEA: Low-Resource Safety Alignment for Multimodal Large Language Models via Synthetic Embeddings
par: Lu, Weikai, et autres
Publié: (2025)

MaskSQL: Safeguarding Privacy for LLM-Based Text-to-SQL via Abstraction
par: Abedini, Sepideh, et autres
Publié: (2025)

Auto-Tuning Safety Guardrails for Black-Box Large Language Models
par: Abdulkadir, Perry
Publié: (2025)

Safeguarding Large Language Models: A Survey
par: Dong, Yi, et autres
Publié: (2024)

SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustness
par: Huo, Jiahao, et autres
Publié: (2026)

MIA-Tuner: Adapting Large Language Models as Pre-training Text Detector
par: Fu, Wenjie, et autres
Publié: (2024)

Adversary-Aware DPO: Enhancing Safety Alignment in Vision Language Models via Adversarial Training
par: Weng, Fenghua, et autres
Publié: (2025)