{"total":30,"rows":[{"id":"2506.07022","title":"AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraint","created":"2025-06-08","cited_by_count":49,"area":"safety"},{"id":"2511.10222","title":"Speech-Audio Compositional Attacks on Multimodal LLMs and Their Mitigation with SALMONN-Guard","created":"2025-11-13","cited_by_count":13,"area":"safety"},{"id":"2502.09755","title":"Jailbreak Attack Initializations as Extractors of Compliance Directions","created":"2025-02-13","cited_by_count":10,"area":"safety"},{"id":"2512.16602","title":"Refusal Steering: Fine-grained Control over LLM Refusal Behaviour for Sensitive Topics","created":"2025-12-18","cited_by_count":6,"area":"safety"},{"id":"2502.16174","title":"Efficient LLM Moderation with Multi-Layer Latent Prototypes","created":"2025-02-22","cited_by_count":4,"area":"safety"},{"id":"2604.08169","title":"Activation Steering for Aligned Open-ended Generation without Sacrificing Coherence","created":"2026-04-09","cited_by_count":4,"area":"safety"},{"id":"2603.14278","title":"Activation Surgery: Jailbreaking White-box LLMs without Touching the Prompt","created":"2026-03-15","cited_by_count":2,"area":"safety"},{"id":"2605.12765","title":"Inference-Time Machine Unlearning via Gated Activation Redirection","created":"2026-05-12","cited_by_count":2,"area":"safety"},{"id":"2605.31293","title":"Divergence Decoding: Inference-Time Unlearning via Auxiliary Models","created":"2026-05-29","cited_by_count":2,"area":"safety"},{"id":"2607.19806","title":"OPIUM: Mitigating Steering Externalities and Over-Refusal via Dual Objective Latent Optimization","created":"2026-07-22","cited_by_count":2,"area":"safety"},{"id":"2506.01770","title":"ReGA: Model-Based Safeguard for LLMs via Representation-Guided Abstraction","created":"2025-06-02","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":"2608.28233","title":"REINS: Refusal-Enhanced Inhibitory Steering with Sparse Autoencoder Features","created":"2026-08-28","cited_by_count":1,"area":"safety"},{"id":"2505.18672","title":"Concept Concentration for Faithful Representation Intervention","created":"2025-05-24","cited_by_count":0,"area":"safety"},{"id":"2601.19487","title":"LLM-VA: Resolving the Jailbreak-Overrefusal Trade-off via Vector Alignment","created":"2026-01-27","cited_by_count":0,"area":"safety"},{"id":"2602.20102","title":"BarrierSteer: LLM Safety via Learning Barrier Steering","created":"2026-02-23","cited_by_count":0,"area":"safety"},{"id":"2604.28129","title":"Latent Adversarial Detection: Adaptive Probing of LLM Activations for Multi-Turn Attack Detection","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":"2605.24552","title":"Ellipsoid Control: A White-list Jailbreak Defense via Benign Latent Modeling","created":"2026-05-23","cited_by_count":0,"area":"safety"},{"id":"2607.22595","title":"xMIx: High-Performance Serving-Time Platform for Mechanistic Interpretability Apps","created":"2026-06-11","cited_by_count":0,"area":"safety"},{"id":"2608.02058","title":"SCOPE: Entanglement Frontier Escape for Source-Free Class Unlearning","created":"2026-08-03","cited_by_count":0,"area":"safety"},{"id":"2609.13478","title":"BadEngram: Backdoor Attack on Gated Memory Components in LLMs","created":"2026-09-11","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":"2609.34475","title":"Causal Routing for Unlearning","created":"2026-09-28","cited_by_count":0,"area":"safety"},{"id":"2609.34514","title":"How to Tame a Multi-Headed Hydra? Adaptive Multi-Category Safety Steering for Large Language Models","created":"2026-09-28","cited_by_count":0,"area":"safety"},{"id":"2609.36657","title":"Constitutional adapters: Inference-time interventions for misalignment and misuse","created":"2026-09-29","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":"2610.04470","title":"Reactivating Alignment: Defending LLMs from Jailbreaks via Intention-Aware Input-Output Matching","created":"2026-10-03","cited_by_count":0,"area":"safety"},{"id":"2610.10276","title":"PatchBench: Measuring Collateral Damage in Activation Patching","created":"2026-10-07","cited_by_count":0,"area":"safety"},{"id":"2610.15365","title":"TRACE: Temporal Risk Assessment and Concept Editing for Safer Language Generation","created":"2026-10-11","cited_by_count":0,"area":"safety"}],"next_cursor":null}