Image de couverture de livre

Silenced Biases: The Dark Side LLMs Learned to Refuse

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Himelstein, Rom, LeVi, Amit, Youngmann, Brit, Nemcovsky, Yaniv, Mendelson, Avi
Format:	Preprint
Publié:	2025
Sujets:	Computation and Language Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Silent Tokens, Loud Effects: Padding in LLMs
par: Himelstein, Rom, et autres
Publié: (2025)

Step-Wise Refusal Dynamics in Autoregressive and Diffusion Language Models
par: Rahimi, Eliron, et autres
Publié: (2026)

Jailbreak Attack Initializations as Extractors of Compliance Directions
par: Levi, Amit, et autres
Publié: (2025)

You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations
par: LeVi, Amit, et autres
Publié: (2025)

Representing LLMs in Prompt Semantic Task Space
par: Kashani, Idan, et autres
Publié: (2025)

Sparse patches adversarial attacks via extrapolating point-wise information
par: Nemcovsky, Yaniv, et autres
Publié: (2024)

REMIND: Input Loss Landscapes Reveal Residual Memorization in Post-Unlearning LLMs
par: Cohen, Liran, et autres
Publié: (2025)

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

Hysteresis Activation Function for Efficient Inference
par: Kimhi, Moshe, et autres
Publié: (2024)

Leveraging NTPs for Efficient Hallucination Detection in VLMs
par: Azachi, Ofir, et autres
Publié: (2025)

Extracting Recurring Vulnerabilities from Black-Box LLM-Generated Software
par: Kordonsky, Tomer, et autres
Publié: (2026)

The Dark Side of the Language: Pre-trained Transformers in the DarkNet
par: Ranaldi, Leonardo, et autres
Publié: (2022)

Training a Bilingual Language Model by Mapping Tokens onto a Shared Character Space
par: Rom, Aviad, et autres
Publié: (2024)

Programming Refusal with Conditional Activation Steering
par: Lee, Bruce W., et autres
Publié: (2024)

Does Refusal Training in LLMs Generalize to the Past Tense?
par: Andriushchenko, Maksym, et autres
Publié: (2024)

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

Prompt Repetition Improves Non-Reasoning LLMs
par: Leviathan, Yaniv, et autres
Publié: (2025)

Learning a Continue-Thinking Token for Enhanced Test-Time Scaling
par: Ringel, Liran, et autres
Publié: (2025)

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

Latent Adversarial Training Improves the Representation of Refusal
par: Abbas, Alexandra, et autres
Publié: (2025)

TOFU: A Task of Fictitious Unlearning for LLMs
par: Maini, Pratyush, et autres
Publié: (2024)

Silence the Judge: Reinforcement Learning with Self-Verifier via Latent Geometric Clustering
par: Zhang, Nonghai, et autres
Publié: (2026)

When Should LLMs Be Less Specific? Selective Abstraction for Reliable Long-Form Text Generation
par: Goren, Shani, et autres
Publié: (2026)

How Post-Training Reshapes LLMs: A Mechanistic View on Knowledge, Truthfulness, Refusal, and Confidence
par: Du, Hongzhe, et autres
Publié: (2025)

A Course Correction in Steerability Evaluation: Revealing Miscalibration and Side Effects in LLMs
par: Chang, Trenton, et autres
Publié: (2025)

Exploiting Synergistic Cognitive Biases to Bypass Safety in LLMs
par: Yang, Xikang, et autres
Publié: (2025)

The Dark Side of ChatGPT: Legal and Ethical Challenges from Stochastic Parrots and Hallucination
par: Li, Zihao
Publié: (2023)

Poison Once, Refuse Forever: Weaponizing Alignment for Injecting Bias in LLMs
par: Mamun, Md Abdullah Al, et autres
Publié: (2025)

Where Do Reasoning Models Refuse?
par: Yamaguchi, Kureha, et autres
Publié: (2025)

Random Initialization of Gated Sparse Adapters
par: Retault, Vi, et autres
Publié: (2025)

The Impossibility of Fair LLMs
par: Anthis, Jacy, et autres
Publié: (2024)

MetaFaith: Faithful Natural Language Uncertainty Expression in LLMs
par: Liu, Gabrielle Kaili-May, et autres
Publié: (2025)

Detection Is Cheap, Routing Is Learned: Why Refusal-Based Alignment Evaluation Fails
par: Frank, Gregory N.
Publié: (2026)

Tracing the Dynamics of Refusal: Exploiting Latent Refusal Trajectories for Robust Jailbreak Detection
par: Hu, Xulin, et autres
Publié: (2026)

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

Learning to Generate Instruction Tuning Datasets for Zero-Shot Task Adaptation
par: Nayak, Nihal V., et autres
Publié: (2024)

Dynamic Adversarial Fine-Tuning Reorganizes Refusal Geometry
par: Lan, Wenhao, et autres
Publié: (2026)

Refusal in Language Models Is Mediated by a Single Direction
par: Arditi, Andy, et autres
Publié: (2024)

Beyond Forgetting: Machine Unlearning Elicits Controllable Side Behaviors and Capabilities
par: Dang, Tien, et autres
Publié: (2026)

ABCD: All Biases Come Disguised
par: Nowak, Mateusz, et autres
Publié: (2026)