Buchumschlag

LSSF: Safety Alignment for Large Language Models through Low-Rank Safety Subspace Fusion

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Zhou, Guanghao, Qiu, Panjia, Chen, Cen, Li, Hongyu, Chu, Mingyuan, Zhang, Xin, Zhou, Jun
Format:	Preprint
Veröffentlicht:	2026
Schlagworte:	Computers and Society Artificial Intelligence
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

CCJA: Context-Coherent Jailbreak Attack for Aligned Large Language Models
von: Zhou, Guanghao, et al.
Veröffentlicht: (2025)

Reinforced MLLM: A Survey on RL-Based Reasoning in Multimodal Large Language Models
von: Zhou, Guanghao, et al.
Veröffentlicht: (2025)

Evaluating Psychological Safety of Large Language Models
von: Li, Xingxuan, et al.
Veröffentlicht: (2022)

Foundational Challenges in Assuring Alignment and Safety of Large Language Models
von: Anwar, Usman, et al.
Veröffentlicht: (2024)

Simple Role Assignment is Extraordinarily Effective for Safety Alignment
von: Ziheng, Zhou, et al.
Veröffentlicht: (2026)

The Hidden Risks of Large Reasoning Models: A Safety Assessment of R1
von: Zhou, Kaiwen, et al.
Veröffentlicht: (2025)

Defining and Evaluating Physical Safety for Large Language Models
von: Tang, Yung-Chen, et al.
Veröffentlicht: (2024)

Interpretable Safety Alignment via SAE-Constructed Low-Rank Subspace Adaptation
von: Wang, Dianyun, et al.
Veröffentlicht: (2025)

How Alignment and Jailbreak Work: Explain LLM Safety through Intermediate Hidden States
von: Zhou, Zhenhong, et al.
Veröffentlicht: (2024)

Benchmarking the Safety of Large Language Models for Robotic Health Attendant Control
von: Nakao, Mahiro, et al.
Veröffentlicht: (2026)

Balancing Safety and Helpfulness in Healthcare AI Assistants through Iterative Preference Alignment
von: Nghiem, Huy, et al.
Veröffentlicht: (2025)

Ensuring Safety and Trust: Analyzing the Risks of Large Language Models in Medicine
von: Yang, Yifan, et al.
Veröffentlicht: (2024)

LocalValueBench: A Collaboratively Built and Extensible Benchmark for Evaluating Localized Value Alignment and Ethical Safety in Large Language Models
von: Meadows, Gwenyth Isobel, et al.
Veröffentlicht: (2024)

LLM Safety Alignment is Divergence Estimation in Disguise
von: Haldar, Rajdeep, et al.
Veröffentlicht: (2025)

Phare: A Safety Probe for Large Language Models
von: Jeune, Pierre Le, et al.
Veröffentlicht: (2025)

Evaluating Interactive Reasoning in Large Language Models: A Hierarchical Benchmark with Executable Games
von: Fan, Mingyuan, et al.
Veröffentlicht: (2026)

SafeRBench: Dissecting the Reasoning Safety of Large Language Models
von: Gao, Xin, et al.
Veröffentlicht: (2025)

Wide Reflective Equilibrium in LLM Alignment: Bridging Moral Epistemology and AI Safety
von: Brophy, Matthew
Veröffentlicht: (2025)

On Fairness of Low-Rank Adaptation of Large Models
von: Ding, Zhoujie, et al.
Veröffentlicht: (2024)

Case-based Reasoning Augmented Large Language Model Framework for Decision Making in Realistic Safety-Critical Driving Scenarios
von: Gan, Wenbin, et al.
Veröffentlicht: (2025)

Information Suppression in Large Language Models: Auditing, Quantifying, and Characterizing Censorship in DeepSeek
von: Qiu, Peiran, et al.
Veröffentlicht: (2025)

Alignment as Iatrogenesis: Pastoral Power, Collective Pathology, and the Structural Limits of Monolingual Safety Evaluation
von: Fukui, Hiroki
Veröffentlicht: (2026)

Understanding Public Safety Trends in Calgary through data mining
von: Dewis, Zack, et al.
Veröffentlicht: (2024)

AI Safety is Stuck in Technical Terms -- A System Safety Response to the International AI Safety Report
von: Dobbe, Roel
Veröffentlicht: (2025)

Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models
von: Duan, Ranjie, et al.
Veröffentlicht: (2025)

The Violation State: Safety State Persistence in a Multimodal Language Model Interface
von: DeVilling, Bentley
Veröffentlicht: (2025)

Safety Cases: A Scalable Approach to Frontier AI Safety
von: Hilton, Benjamin, et al.
Veröffentlicht: (2025)

Safety Cases: How to Justify the Safety of Advanced AI Systems
von: Clymer, Joshua, et al.
Veröffentlicht: (2024)

PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approach
von: Sehwag, Udari Madhushani, et al.
Veröffentlicht: (2025)

Superficial Safety Alignment Hypothesis
von: Li, Jianwei, et al.
Veröffentlicht: (2024)

Urban Safety Perception Through the Lens of Large Multimodal Models: A Persona-based Approach
von: Beneduce, Ciro, et al.
Veröffentlicht: (2025)

Alignment and Safety in Large Language Models: Safety Mechanisms, Training Paradigms, and Emerging Challenges
von: Lu, Haoran, et al.
Veröffentlicht: (2025)

LLM Safety for Children
von: Rath, Prasanjit, et al.
Veröffentlicht: (2025)

Intelligent Computing Social Modeling and Methodological Innovations in Political Science in the Era of Large Language Models
von: Wang, Zhenyu, et al.
Veröffentlicht: (2024)

Towards Context-Invariant Safety Alignment for Large Language Models
von: Wang, Yixu, et al.
Veröffentlicht: (2026)

International Agreements on AI Safety: Review and Recommendations for a Conditional AI Safety Treaty
von: Scholefield, Rebecca, et al.
Veröffentlicht: (2025)

Reducing Large Language Model Safety Risks in Women's Health using Semantic Entropy
von: Penny-Dimri, Jahan C., et al.
Veröffentlicht: (2025)

Disentangling AI Alignment: A Structured Taxonomy Beyond Safety and Ethics
von: Baum, Kevin
Veröffentlicht: (2025)

Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning
von: Alssum, Lama, et al.
Veröffentlicht: (2025)

Mitigating Gambling-Like Risk-Taking Behaviors in Large Language Models: A Behavioral Economics Approach to AI Safety
von: Du, Y.
Veröffentlicht: (2025)