Image de couverture de livre

Surgical Refusal Ablation: Disentangling Safety from Intelligence via Concept-Guided Spectral Cleaning

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteur principal:	Cristofano, Tony
Format:	Preprint
Publié:	2026
Sujets:	Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Universal Refusal Circuits Across LLMs: Cross-Model Transfer via Trajectory Replay and Concept-Basis Reconstruction
par: Cristofano, Tony
Publié: (2026)

Surgical, Cheap, and Flexible: Mitigating False Refusal in Language Models via Single Vector Ablation
par: Wang, Xinpeng, et autres
Publié: (2024)

Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training
par: Yuan, Youliang, et autres
Publié: (2024)

Applying Refusal-Vector Ablation to Llama 3.1 70B Agents
par: Lermen, Simon, et autres
Publié: (2024)

Think Before Refusal : Triggering Safety Reflection in LLMs to Mitigate False Refusal Behavior
par: Si, Shengyun, et autres
Publié: (2025)

Anchoring Refusal Direction: Mitigating Safety Risks in Tuning via Projection Constraint
par: Du, Yanrui, et autres
Publié: (2025)

Safety-Aligned Weights Are Not Enough: Refusal-Teacher-Guided Finetuning Enhances Safety and Downstream Performance under Harmful Finetuning Attacks
par: Ham, Seokil, et autres
Publié: (2025)

Refusal Direction is Universal Across Safety-Aligned Languages
par: Wang, Xinpeng, et autres
Publié: (2025)

Should LLM Safety Be More Than Refusing Harmful Instructions?
par: Maskey, Utsav, et autres
Publié: (2025)

RepIt: Steering Language Models with Concept-Specific Refusal Vectors
par: Siu, Vincent, et autres
Publié: (2025)

DualEdit: Mitigating Safety Fallback in LLM Backdoor Editing via Affirmation-Refusal Regulation
par: Jiang, Houcheng, et autres
Publié: (2025)

Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety
par: Zhang, Yuyou, et autres
Publié: (2025)

When Safety Blocks Sense: Measuring Semantic Confusion in LLM Refusals
par: Anonto, Riad Ahmed, et autres
Publié: (2025)

The Geometry of Refusal in Large Language Models: Concept Cones and Representational Independence
par: Wollschläger, Tom, et autres
Publié: (2025)

From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Training
par: Yuan, Yuan, et autres
Publié: (2025)

WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs
par: Han, Seungju, et autres
Publié: (2024)

FalseReject: A Resource for Improving Contextual Safety and Mitigating Over-Refusals in LLMs via Structured Reasoning
par: Zhang, Zhehao, et autres
Publié: (2025)

RefusalGuard: Geometry-Preserving Fine-Tuning for Safety in LLMs
par: Asif, Sadia, et autres
Publié: (2026)

Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models
par: Duan, Ranjie, et autres
Publié: (2025)

Do Language Models Know When They'll Refuse? Probing Introspective Awareness of Safety Boundaries
par: Gondil, Tanay
Publié: (2026)

Over-Refusal and Representation Subspaces: A Mechanistic Analysis of Task-Conditioned Refusal in Aligned LLMs
par: Maskey, Utsav, et autres
Publié: (2026)

Beyond Over-Refusal: Scenario-Based Diagnostics and Post-Hoc Mitigation for Exaggerated Refusals in LLMs
par: Yuan, Shuzhou, et autres
Publié: (2025)

When Refusals Fail: Unstable Safety Mechanisms in Long-Context LLM Agents
par: Hadeliya, Tsimur, et autres
Publié: (2025)

From Threat to Tool: Leveraging Refusal-Aware Injection Attacks for Safety Alignment
par: Chae, Kyubyung, et autres
Publié: (2025)

ConceptGuard: Neuro-Symbolic Safety Guardrails via Sparse Interpretable Jailbreak Concepts
par: Aswal, Darpan, et autres
Publié: (2025)

Refusal Tokens: A Simple Way to Calibrate Refusals in Large Language Models
par: Jain, Neel, et autres
Publié: (2024)

ORFuzz: Fuzzing the "Other Side" of LLM Safety -- Testing Over-Refusal
par: Zhang, Haonan, et autres
Publié: (2025)

Cyclic Ablation: Testing Concept Localization against Functional Regeneration in AI
par: Kapelko, Eduard
Publié: (2025)

LLMs Encode Harmfulness and Refusal Separately
par: Zhao, Jiachen, et autres
Publié: (2025)

Refusal Steering: Fine-grained Control over LLM Refusal Behaviour for Sensitive Topics
par: García-Ferrero, Iker, et autres
Publié: (2025)

Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuning
par: Casademunt, Helena, et autres
Publié: (2025)

RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models
par: Muhamed, Aashiq, et autres
Publié: (2025)

Refusal in LLMs is an Affine Function
par: Marshall, Thomas, et autres
Publié: (2024)

Can LLMs Refuse Questions They Do Not Know? Measuring Knowledge-Aware Refusal in Factual Tasks
par: Pan, Wenbo, et autres
Publié: (2025)

Discern Truth from Falsehood: Reducing Over-Refusal via Contrastive Refinement
par: Lu, Yuxiao, et autres
Publié: (2026)

Don't Say No: Jailbreaking LLM by Suppressing Refusal
par: Zhou, Yukai, et autres
Publié: (2024)

Understanding Refusal in Language Models with Sparse Autoencoders
par: Yeo, Wei Jie, et autres
Publié: (2025)

Towards Understanding and Improving Refusal in Compressed Models via Mechanistic Interpretability
par: Chhabra, Vishnu Kabir, et autres
Publié: (2025)

Latent Concept Disentanglement in Transformer-based Language Models
par: Hong, Guan Zhe, et autres
Publié: (2025)

Characterizing Selective Refusal Bias in Large Language Models
par: Khorramrouz, Adel, et autres
Publié: (2025)