Cover Image

Knowing without Acting: The Disentangled Geometry of Safety Mechanisms in Large Language Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wu, Jinman, Xie, Yi, Lin, Shen, Zhao, Shiqian, Chen, Xiaofeng
Format:	Preprint
Published:	2026
Subjects:	Cryptography and Security Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Depth Charge: Jailbreak Large Language Models from Deep Safety Attention Heads
by: Wu, Jinman, et al.
Published: (2026)

SAME: Sample Reconstruction against Model Extraction Attacks
by: Xie, Yi, et al.
Published: (2023)

Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning
by: Liu, Guozhi, et al.
Published: (2025)

UpSafe$^\circ$C: Upcycling for Controllable Safety in Large Language Models
by: Sun, Yuhao, et al.
Published: (2025)

Understanding the Effects of Safety Unalignment on Large Language Models
by: Halloran, John T.
Published: (2026)

When Safety Geometry Collapses: Fine-Tuning Vulnerabilities in Agentic Guard Models
by: Hossain, Ismail, et al.
Published: (2026)

Safety Layers in Aligned Large Language Models: The Key to LLM Security
by: Li, Shen, et al.
Published: (2024)

DMark: Order-Agnostic Watermarking for Diffusion Large Language Models
by: Wu, Linyu, et al.
Published: (2025)

Lifelong Safety Alignment for Language Models
by: Wang, Haoyu, et al.
Published: (2025)

AISA: Awakening Intrinsic Safety Awareness in Large Language Models against Jailbreak Attacks
by: Song, Weiming, et al.
Published: (2026)

Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable
by: Huang, Tiansheng, et al.
Published: (2025)

KnowPhish: Large Language Models Meet Multimodal Knowledge Graphs for Enhancing Reference-Based Phishing Detection
by: Li, Yuexin, et al.
Published: (2024)

Probing the Safety Response Boundary of Large Language Models via Unsafe Decoding Path Generation
by: Wang, Haoyu, et al.
Published: (2024)

A Survey on Model Extraction Attacks and Defenses for Large Language Models
by: Zhao, Kaixiang, et al.
Published: (2025)

Prompt, Divide, and Conquer: Bypassing Large Language Model Safety Filters via Segmented and Distributed Prompt Processing
by: Wahréus, Johan, et al.
Published: (2025)

Reconstructing Training Data from Adapter-based Federated Large Language Models
by: Chen, Silong, et al.
Published: (2026)

Model-based Large Language Model Customization as Service
by: Wu, Zhaomin, et al.
Published: (2024)

Permissioned LLMs: Enforcing Access Control in Large Language Models
by: Jayaraman, Bargav, et al.
Published: (2025)

A Survey on Data Security in Large Language Models
by: Chen, Kang, et al.
Published: (2025)

Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction
by: Guo, Jiahe, et al.
Published: (2026)

Privacy-Preserving Large Language Models: Mechanisms, Applications, and Future Directions
by: Zhao, Guoshenghui, et al.
Published: (2024)

SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models
by: Li, Lijun, et al.
Published: (2024)

On the Role of Attention Heads in Large Language Model Safety
by: Zhou, Zhenhong, et al.
Published: (2024)

Large Language Model Empowered Privacy-Protected Framework for PHI Annotation in Clinical Notes
by: Wu, Guanchen, et al.
Published: (2025)

SafeMLRM: Demystifying Safety in Multi-modal Large Reasoning Models
by: Fang, Junfeng, et al.
Published: (2025)

Exploring the Secondary Risks of Large Language Models
by: Chen, Jiawei, et al.
Published: (2025)

TrafficLLM: Enhancing Large Language Models for Network Traffic Analysis with Generic Traffic Representation
by: Cui, Tianyu, et al.
Published: (2025)

Jailbreaking and Mitigation of Vulnerabilities in Large Language Models
by: Peng, Benji, et al.
Published: (2024)

Probing the Robustness of Large Language Models Safety to Latent Perturbations
by: Gu, Tianle, et al.
Published: (2025)

Ablating Safety: Mechanisms for Removing Alignment in Language Models for Security Applications
by: David, Isaac, et al.
Published: (2026)

Oblivionis: A Lightweight Learning and Unlearning Framework for Federated Large Language Models
by: Zhang, Fuyao, et al.
Published: (2025)

DR-Encoder: Encode Low-rank Gradients with Random Prior for Large Language Models Differentially Privately
by: Wu, Huiwen, et al.
Published: (2024)

Watermark Stealing in Large Language Models
by: Jovanović, Nikola, et al.
Published: (2024)

Privacy Auditing of Large Language Models
by: Panda, Ashwinee, et al.
Published: (2025)

NeuroBreak: Unveil Internal Jailbreak Mechanisms in Large Language Models
by: Zhang, Chuhan, et al.
Published: (2025)

SAID: Safety-Aware Intent Defense via Prefix Probing for Large Language Models
by: Chen, Yulong, et al.
Published: (2025)

PrivTune: Efficient and Privacy-Preserving Fine-Tuning of Large Language Models via Device-Cloud Collaboration
by: Liu, Yi, et al.
Published: (2025)

Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment
by: Li, Yuxi, et al.
Published: (2024)

FRAMU: Attention-based Machine Unlearning using Federated Reinforcement Learning
by: Shaik, Thanveer, et al.
Published: (2023)

Investigating the Impact of Quantization Methods on the Safety and Reliability of Large Language Models
by: Kharinaev, Artyom, et al.
Published: (2025)