Imagen de Portada

Provable Defense Framework for LLM Jailbreaks via Noise-Augumented Alignment

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Cheng, Zehua, Yang, Jianwei, Dai, Wei, Sun, Jiahao
Formato:	Preprint
Publicado:	2026
Materias:	Computation and Language Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

CircuitSynth: Reliable Synthetic Data Generation
por: Cheng, Zehua, et al.
Publicado: (2026)

CasualSynth: Generating Structurally Sound Synthetic Data
por: Cheng, Zehua, et al.
Publicado: (2026)

Efficient LLM-Jailbreaking via Multimodal-LLM Jailbreak
por: Ji, Haoxuan, et al.
Publicado: (2024)

AttentionDefense: Leveraging System Prompt Attention for Explainable Defense Against Novel Jailbreaks
por: Siska, Charlotte, et al.
Publicado: (2025)

SelfPrompt: Autonomously Evaluating LLM Robustness via Domain-Constrained Knowledge Guidelines and Refined Adversarial Prompts
por: Pei, Aihua, et al.
Publicado: (2024)

THRD: A Training-Free Multi-Turn Defense Framework for Jailbreak Attacks on Large Language Models
por: Ma, Zhiqing, et al.
Publicado: (2026)

When "Competency" in Reasoning Opens the Door to Vulnerability: Jailbreaking LLMs via Novel Complex Ciphers
por: Handa, Divij, et al.
Publicado: (2024)

How Jailbreak Defenses Work and Ensemble? A Mechanistic Investigation
por: Long, Zhuohang, et al.
Publicado: (2025)

A Graph-Enhanced Defense Framework for Explainable Fake News Detection with LLM
por: Wang, Bo, et al.
Publicado: (2026)

Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation
por: Sun, Chenkai, et al.
Publicado: (2025)

Efficient Knowledge Infusion via KG-LLM Alignment
por: Jiang, Zhouyu, et al.
Publicado: (2024)

SafeMCP: Proactive Power Regulation for LLM Agent Defense via Environment-Grounded Look-Ahead Reasoning
por: Wang, Lichao, et al.
Publicado: (2026)

EasyJailbreak: A Unified Framework for Jailbreaking Large Language Models
por: Zhou, Weikang, et al.
Publicado: (2024)

Adversarial Preference Optimization: Enhancing Your Alignment via RM-LLM Game
por: Cheng, Pengyu, et al.
Publicado: (2023)

ChunkLLM: A Lightweight Pluggable Framework for Accelerating LLMs Inference
por: Ouyang, Haojie, et al.
Publicado: (2025)

The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis
por: Wei, Zihao, et al.
Publicado: (2025)

Test-Time Immunization: A Universal Defense Framework Against Jailbreaks for (Multimodal) Large Language Models
por: Yu, Yongcan, et al.
Publicado: (2025)

Societal Alignment Frameworks Can Improve LLM Alignment
por: Stańczak, Karolina, et al.
Publicado: (2025)

SATA: A Paradigm for LLM Jailbreak via Simple Assistive Task Linkage
por: Dong, Xiaoning, et al.
Publicado: (2024)

Jailbreaking to Jailbreak
por: Kritz, Jeremy, et al.
Publicado: (2025)

Self-Guided Defense: Adaptive Safety Alignment for Reasoning Models via Synthesized Guidelines
por: Wang, Yuhang, et al.
Publicado: (2025)

Zero-Shot Defense Against Toxic Images via Inherent Multimodal Alignment in LVLMs
por: Zhao, Wei, et al.
Publicado: (2025)

Scalable Defense against In-the-wild Jailbreaking Attacks with Safety Context Retrieval
por: Chen, Taiye, et al.
Publicado: (2025)

FRACTURED-SORRY-Bench: Framework for Revealing Attacks in Conversational Turns Undermining Refusal Efficacy and Defenses over SORRY-Bench (Automated Multi-shot Jailbreaks)
por: Priyanshu, Aman, et al.
Publicado: (2024)

PMMT: Preference Alignment in Multilingual Machine Translation via LLM Distillation
por: Sun, Shuqiao, et al.
Publicado: (2024)

Attention Slipping: A Mechanistic Understanding of Jailbreak Attacks and Defenses in LLMs
por: Hu, Xiaomeng, et al.
Publicado: (2025)

ShieldLearner: A New Paradigm for Jailbreak Attack Defense in LLMs
por: Ni, Ziyi, et al.
Publicado: (2025)

Layer-Level Self-Exposure and Patch: Affirmative Token Mitigation for Jailbreak Attack Defense
por: Ouyang, Yang, et al.
Publicado: (2025)

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

Beyond Single-Point Judgment: Distribution Alignment for LLM-as-a-Judge
por: Chen, Luyu, et al.
Publicado: (2025)

Advancing Jailbreak Strategies: A Hybrid Approach to Exploiting LLM Vulnerabilities and Bypassing Modern Defenses
por: Ahmed, Mohamed, et al.
Publicado: (2025)

Diverse Human Value Alignment for Large Language Models via Ethical Reasoning
por: Wang, Jiahao, et al.
Publicado: (2025)

Human-Instruction-Free LLM Self-Alignment with Limited Samples
por: Guo, Hongyi, et al.
Publicado: (2024)

Jailbreak Attacks and Defenses Against Large Language Models: A Survey
por: Yi, Sibo, et al.
Publicado: (2024)

Persona-Aware Alignment Framework for Personalized Dialogue Generation
por: Li, Guanrong, et al.
Publicado: (2025)

Na'vi or Knave: Jailbreaking Language Models via Metaphorical Avatars
por: Yan, Yu, et al.
Publicado: (2024)

Defensive Dual Masking for Robust Adversarial Defense
por: Yang, Wangli, et al.
Publicado: (2024)

Many-Turn Jailbreaking
por: Yang, Xianjun, et al.
Publicado: (2025)

KGPA: Robustness Evaluation for Large Language Models via Cross-Domain Knowledge Graphs
por: Pei, Aihua, et al.
Publicado: (2024)

Collab: Controlled Decoding using Mixture of Agents for LLM Alignment
por: Chakraborty, Souradip, et al.
Publicado: (2025)