{"total":55,"rows":[{"id":"2406.11717","title":"Refusal in Language Models Is Mediated by a Single Direction","created":"2024-06-17","cited_by_count":1032,"area":"safety"},{"id":"2407.15549","title":"Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs","created":"2024-07-22","cited_by_count":173,"area":"safety"},{"id":"2410.19278","title":"Applying sparse autoencoders to unlearn knowledge in language models","created":"2024-10-25","cited_by_count":59,"area":"safety"},{"id":"2412.06878","title":"SafeWatch: An Efficient Safety-Policy Following Video Guardrail Model with Transparent Explanations","created":"2024-12-09","cited_by_count":38,"area":"safety"},{"id":"2502.01042","title":"SafeSwitch: Steering Unsafe LLM Behavior via Internal Activation Signals","created":"2025-02-03","cited_by_count":29,"area":"safety"},{"id":"2411.11114","title":"JailbreakLens: Interpreting Jailbreak Mechanism in the Lens of Representation and Circuit","created":"2024-11-17","cited_by_count":22,"area":"safety"},{"id":"2504.08192","title":"SAEs Can Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs","created":"2025-04-11","cited_by_count":22,"area":"safety"},{"id":"2510.14005","title":"PIShield: Detecting Prompt Injection Attacks via Intrinsic LLM Features","created":"2025-10-15","cited_by_count":20,"area":"safety"},{"id":"2509.09708","title":"Beyond I'm Sorry, I Can't: Dissecting Large Language Model Refusal","created":"2025-09-07","cited_by_count":19,"area":"safety"},{"id":"2507.09709","title":"Large Language Models Encode Semantics and Alignment in Linearly Separable Representations","created":"2025-07-13","cited_by_count":17,"area":"safety"},{"id":"2509.26238","title":"Beyond Linear Probes: Dynamic Safety Monitoring for Language Models","created":"2025-09-30","cited_by_count":16,"area":"safety"},{"id":"2505.24445","title":"Learning Safety Constraints for Large Language Models","created":"2025-05-30","cited_by_count":13,"area":"safety"},{"id":"2506.09227","title":"SoK: Machine Unlearning for Large Language Models","created":"2025-06-10","cited_by_count":11,"area":"safety"},{"id":"2508.13650","title":"CRISP: Persistent Concept Unlearning via Sparse Autoencoders","created":"2025-08-19","cited_by_count":11,"area":"safety"},{"id":"2506.05451","title":"Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safety","created":"2025-06-05","cited_by_count":10,"area":"safety"},{"id":"2503.21730","title":"Effective Skill Unlearning through Intervention and Abstention","created":"2025-03-27","cited_by_count":9,"area":"safety"},{"id":"2604.22167","title":"Estimating Tail Risks in Language Model Output Distributions","created":"2026-04-24","cited_by_count":9,"area":"safety"},{"id":"2411.07122","title":"SCAR: Sparse Conditioned Autoencoders for Concept Detection and Steering in LLMs","created":"2024-11-11","cited_by_count":7,"area":"safety"},{"id":"2508.00161","title":"Watch the Weights: Unsupervised monitoring and control of fine-tuned LLMs","created":"2025-07-31","cited_by_count":7,"area":"safety"},{"id":"2508.08211","title":"SAEMark: Steering Personalized Multilingual LLM Watermarks with Sparse Autoencoders","created":"2025-08-11","cited_by_count":7,"area":"safety"},{"id":"2503.11127","title":"Don't Forget It! Conditional Sparse Autoencoder Clamping Works for Unlearning","created":"2025-03-14","cited_by_count":6,"area":"safety"},{"id":"2505.07610","title":"Concept-Level Explainability for Auditing & Steering LLM Responses","created":"2025-05-12","cited_by_count":6,"area":"safety"},{"id":"2602.12418","title":"Sparse Autoencoders are Capable LLM Jailbreak Mitigators","created":"2026-02-12","cited_by_count":6,"area":"safety"},{"id":"2505.20435","title":"The Shape of Adversarial Influence: Characterizing LLM Latent Spaces with Persistent Homology","created":"2025-05-26","cited_by_count":5,"area":"safety"},{"id":"2508.15449","title":"Reliable Unlearning Harmful Information in LLMs with Metamorphosis Representation Projection","created":"2025-08-21","cited_by_count":4,"area":"safety"},{"id":"2508.15847","title":"Mechanistic Exploration of Backdoored Large Language Model Attention Patterns","created":"2025-08-19","cited_by_count":3,"area":"safety"},{"id":"2509.15631","title":"Sparse-Autoencoder-Guided Internal Representation Unlearning for Large Language Models","created":"2025-09-19","cited_by_count":3,"area":"safety"},{"id":"2508.16325","title":"ConceptGuard: Neuro-Symbolic Safety Guardrails via Sparse Interpretable Jailbreak Concepts","created":"2025-08-22","cited_by_count":2,"area":"safety"},{"id":"2605.05443","title":"SLAM: Structural Linguistic Activation Marking for Language Models","created":"2026-05-06","cited_by_count":2,"area":"safety"},{"id":"2410.08069","title":"Unlearning-based Neural Interpretations","created":"2024-10-10","cited_by_count":1,"area":"safety"},{"id":"2508.10404","title":"Layer-Wise Perturbations via Sparse Autoencoders for Adversarial Text Generation","created":"2025-08-14","cited_by_count":1,"area":"safety"},{"id":"2512.06655","title":"Graph-Regularized Sparse Autoencoders for LLM Safety Steering","created":"2025-12-07","cited_by_count":1,"area":"safety"},{"id":"2608.07430","title":"Diffusion LLMs as Targets and Adversaries: Mechanistic Safety Exploits","created":"2026-08-07","cited_by_count":1,"area":"safety"},{"id":"2510.12672","title":"Keep Calm and Avoid Harmful Content: Concept Alignment and Latent Manipulation Towards Safer Answers","created":"2025-10-14","cited_by_count":0,"area":"safety"},{"id":"2601.12937","title":"On the Evidentiary Limits of Membership Inference for Copyright Auditing","created":"2026-01-19","cited_by_count":0,"area":"safety"},{"id":"2602.08877","title":"Stress-Testing Alignment Audits With Prompt-Level Strategic Deception","created":"2026-02-09","cited_by_count":0,"area":"safety"},{"id":"2604.18756","title":"Towards Understanding the Robustness of Sparse Autoencoders","created":"2026-04-20","cited_by_count":0,"area":"safety"},{"id":"2604.20945","title":"Breaking Bad: Interpretability-Based Safety Audits of State-of-the-Art LLMs","created":"2026-04-22","cited_by_count":0,"area":"safety"},{"id":"2604.23130","title":"From Concept-Aligned Tokens to Vulnerable Features: Mechanistic Localization of Jailbreaks","created":"2026-04-25","cited_by_count":0,"area":"safety"},{"id":"2605.00123","title":"Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models","created":"2026-04-30","cited_by_count":0,"area":"safety"},{"id":"2605.24535","title":"Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation","created":"2026-05-23","cited_by_count":0,"area":"safety"},{"id":"2606.29646","title":"Fuzzing Large Language Models to Elicit Hidden Behaviours","created":"2026-06-28","cited_by_count":0,"area":"safety"},{"id":"2607.07903","title":"Mechanistic Interpretability of LLM Jailbreaks via Internal Attribution Graphs","created":"2026-07-08","cited_by_count":0,"area":"safety"},{"id":"2607.15218","title":"When Words Are Safe But Actions Kill: Probing Physical Jailbreak Beyond Textual Jailbreak in Hidden-State Risk Space","created":"2026-07-16","cited_by_count":0,"area":"safety"},{"id":"2607.19366","title":"Geometry-Guided Constraint Learning for LLM Safety Classification","created":"2026-06-09","cited_by_count":0,"area":"safety"},{"id":"2608.18089","title":"Latent Space Refusal Anchoring for Low-Resource African Languages: Mechanistic Safety Recovery Without Retraining","created":"2026-06-05","cited_by_count":0,"area":"safety"},{"id":"2609.07746","title":"LLM Forensics: Where Do Backdoors Hide? Localizing and Controlling Trigger Mechanisms with Sparse Autoencoders","created":"2026-09-07","cited_by_count":0,"area":"safety"},{"id":"2609.14861","title":"Semantic Fibers and Cross-Gram Interference: A Calculus of Safety Drift in Overcomplete Representations","created":"2026-09-14","cited_by_count":0,"area":"safety"},{"id":"2609.16229","title":"Test-Time Unlearning via Sparse Autoencoder","created":"2026-09-14","cited_by_count":0,"area":"safety"},{"id":"2609.20129","title":"Local Sparsity Enables Unsupervised LLM Safety Detection","created":"2026-09-17","cited_by_count":0,"area":"safety"},{"id":"2609.31056","title":"Neuralyzing the Trace: Selective Representation-Level Unlearning with Contrastive Sparse Autoencoders","created":"2026-09-25","cited_by_count":0,"area":"safety"},{"id":"2609.35544","title":"Less Sycophancy, Stronger Refusal? Lessons for AI Safety from Mechanistic Interpretability","created":"2026-09-28","cited_by_count":0,"area":"safety"},{"id":"2610.04245","title":"On the Steering Dimensionality of Refusal in Language Models","created":"2026-10-03","cited_by_count":0,"area":"safety"},{"id":"2610.06049","title":"Backdooring Sparse Autoencoders","created":"2026-10-05","cited_by_count":0,"area":"safety"},{"id":"2610.10203","title":"How to train your model organism","created":"2026-10-07","cited_by_count":0,"area":"safety"}],"next_cursor":null}