There Is More to Refusal in Large Language Models than a Single Direction

arXiv:2602.02132 · cs.CL · Submitted 2026-02-02 · Read on arXiv

cs.CL

Submitted: 2026-02-02

Updated: 2026-09-15

Comments: 37 pages. Accepted for publication in the main track of EMNLP 2026. Updated manuscript

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Steering — 48× in this paper · explained in 2 episodes
• Alignment — 15× in this paper · explained in 1 episode
• Ablation — 11× in this paper · explained in 1 episode
• Cosine Similarity — 11× in this paper · explained in 3 episodes
• Accuracy — 10× in this paper · explained in 1 episode
• Interpretability — 7× in this paper · explained in 15 episodes
• Core — 6× in this paper · explained in 2 episodes
• Abliteration — 5× in this paper · explained in 1 episode
• Latent Space — 5× in this paper · explained in 7 episodes
• Activation Addition — 3× in this paper · explained in 1 episode
• Adam — 3× in this paper · explained in 2 episodes
• Benchmark — 3× in this paper · explained in 2 episodes
• Representation Engineering — 3× in this paper · explained in 2 episodes
• Robustness — 3× in this paper · explained in 5 episodes
• Sparse Autoencoders (SAEs) — 3× in this paper · explained in 3 episodes
• Activation Steering — 2× in this paper · explained in 4 episodes
• Reinforcement Learning — 2× in this paper · explained in 4 episodes
• Transformer — 2× in this paper · explained in 2 episodes
• Clustering — 1× in this paper · explained in 2 episodes
• Fidelity — 1× in this paper · explained in 2 episodes
• Fine-Tuning — 1× in this paper · explained in 5 episodes
• Hallucination — 1× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Sycophancy — 1× in this paper · explained in 4 episodes

Sources

• A General Language Assistant as a Laboratory for Alignment
• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Constitutional AI: Harmlessness from AI Feedback
• Safe RLHF: Safe Reinforcement Learning from Human Feedback
• Do I Know This Entity? Knowledge Awareness and Hallucinations in Language Models
• Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned
• Aligning AI With Shared Human Values
• Beyond I'm Sorry, I Can't: Dissecting Large Language Model Refusal
• Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2
• Steering Language Model Refusal with Sparse Autoencoders
• Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoders
• The Geometry of Refusal in Large Language Models: Concept Cones and Representational Independence
• From Instructions to Intrinsic Human Values -- A Survey of Alignment Goals for Big Models
• Understanding Refusal in Language Models with Sparse Autoencoders
• Differentiated Directional Intervention A Framework for Evading LLM Safety Alignment
• Representation Engineering: A Top-Down Approach to AI Transparency

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering