{"total":102,"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":"2210.01504","title":"Knowledge Unlearning for Mitigating Privacy Risks in Language Models","created":"2022-10-04","cited_by_count":437,"area":"safety"},{"id":"2310.02238","title":"Who's Harry Potter? Approximate Unlearning in LLMs","created":"2023-10-03","cited_by_count":381,"area":"safety"},{"id":"2406.04313","title":"Improving Alignment and Robustness with Circuit Breakers","created":"2024-06-06","cited_by_count":322,"area":"safety"},{"id":"2406.07933","title":"Large Language Model Unlearning via Embedding-Corrupted Prompts","created":"2024-06-12","cited_by_count":116,"area":"safety"},{"id":"2305.14710","title":"Instructions as Backdoors: Backdoor Vulnerabilities of Instruction Tuning for Large Language Models","created":"2023-05-24","cited_by_count":110,"area":"safety"},{"id":"2404.18239","title":"SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning","created":"2024-04-28","cited_by_count":104,"area":"safety"},{"id":"2403.03329","title":"Guardrail Baselines for Unlearning in LLMs","created":"2024-03-05","cited_by_count":96,"area":"safety"},{"id":"2405.14577","title":"Representation Noising: A Defence Mechanism Against Harmful Finetuning","created":"2024-05-23","cited_by_count":96,"area":"safety"},{"id":"2406.08607","title":"Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference","created":"2024-06-12","cited_by_count":88,"area":"safety"},{"id":"2406.10890","title":"RWKU: Benchmarking Real-World Knowledge Unlearning for Large Language Models","created":"2024-06-16","cited_by_count":78,"area":"safety"},{"id":"2410.11143","title":"LLM Unlearning via Loss Adjustment with Only Forget Data","created":"2024-10-14","cited_by_count":75,"area":"safety"},{"id":"2510.07192","title":"Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples","created":"2025-10-08","cited_by_count":72,"area":"safety"},{"id":"2410.02879","title":"Position: LLM Unlearning Benchmarks are Weak Measures of Progress","created":"2024-10-03","cited_by_count":63,"area":"safety"},{"id":"2310.15851","title":"Self-Guard: Empower the LLM to Safeguard Itself","created":"2023-10-24","cited_by_count":54,"area":"safety"},{"id":"2405.18540","title":"Learning diverse attacks on large language models for robust red-teaming and safety tuning","created":"2024-05-28","cited_by_count":54,"area":"safety"},{"id":"2404.05880","title":"Eraser: Jailbreaking Defense in Large Language Models via Unlearning Harmful Knowledge","created":"2024-04-08","cited_by_count":51,"area":"safety"},{"id":"2402.10052","title":"UNDIAL: Self-Distillation with Adjusted Logits for Robust Unlearning in Large Language Models","created":"2024-02-15","cited_by_count":50,"area":"safety"},{"id":"2407.20516","title":"Machine Unlearning in Generative AI: A Survey","created":"2024-07-30","cited_by_count":49,"area":"safety"},{"id":"2405.03097","title":"To Each (Textual Sequence) Its Own: Improving Memorized-Data Unlearning in Large Language Models","created":"2024-05-06","cited_by_count":43,"area":"safety"},{"id":"2404.02062","title":"Digital Forgetting in Large Language Models: A Survey of Unlearning Methods","created":"2024-04-02","cited_by_count":40,"area":"safety"},{"id":"2402.05813","title":"Selective Forgetting: Advancing Machine Unlearning Techniques and Evaluation in Language Models","created":"2024-02-08","cited_by_count":39,"area":"safety"},{"id":"2404.11045","title":"Offset Unlearning for Large Language Models","created":"2024-04-17","cited_by_count":36,"area":"safety"},{"id":"2406.15481","title":"Code-Switching Red-Teaming: LLM Evaluation for Safety and Multilingual Understanding","created":"2024-06-17","cited_by_count":34,"area":"safety"},{"id":"2402.15302","title":"How (un)ethical are instruction-centric responses of LLMs? Unveiling the vulnerabilities of safety guardrails to harmful queries","created":"2024-02-23","cited_by_count":27,"area":"safety"},{"id":"2406.16257","title":"Towards Scalable Exact Machine Unlearning Using Parameter-Efficient Fine-Tuning","created":"2024-06-24","cited_by_count":27,"area":"safety"},{"id":"2401.10862","title":"Pruning for Protection: Increasing Jailbreak Resistance in Aligned LLMs Without Fine-Tuning","created":"2024-01-19","cited_by_count":26,"area":"safety"},{"id":"2410.17509","title":"WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models","created":"2024-10-23","cited_by_count":26,"area":"safety"},{"id":"2310.14303","title":"Language Model Unalignment: Parametric Red-Teaming to Expose Hidden Harms and Biases","created":"2023-10-22","cited_by_count":24,"area":"safety"},{"id":"2503.04636","title":"Mark Your LLM: Detecting the Misuse of Open-Source Large Language Models via Watermarking","created":"2025-03-06","cited_by_count":19,"area":"safety"},{"id":"2506.07596","title":"TwinBreak: Jailbreaking LLM Security Alignments based on Twin Prompts","created":"2025-06-09","cited_by_count":19,"area":"safety"},{"id":"2406.11780","title":"Split, Unlearn, Merge: Leveraging Data Attributes for More Effective Unlearning in LLMs","created":"2024-06-17","cited_by_count":18,"area":"safety"},{"id":"2505.13312","title":"GUARD: Generation-time LLM Unlearning via Adaptive Restriction and Detection","created":"2025-05-19","cited_by_count":17,"area":"safety"},{"id":"2405.16720","title":"Large Scale Knowledge Washing","created":"2024-05-26","cited_by_count":16,"area":"safety"},{"id":"2311.09816","title":"Downstream Trade-offs of a Family of Text Watermarks","created":"2023-11-16","cited_by_count":14,"area":"safety"},{"id":"2504.07467","title":"Defense against Prompt Injection Attacks via Mixture of Encodings","created":"2025-04-10","cited_by_count":14,"area":"safety"},{"id":"2407.11282","title":"Uncertainty is Fragile: Manipulating Uncertainty in Large Language Models","created":"2024-07-15","cited_by_count":13,"area":"safety"},{"id":"2407.01376","title":"Badllama 3: removing safety finetuning from Llama 3 in minutes","created":"2024-07-01","cited_by_count":11,"area":"safety"},{"id":"2406.12329","title":"Opt-Out: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport","created":"2024-06-18","cited_by_count":10,"area":"safety"},{"id":"2406.14091","title":"Protecting Privacy Through Approximating Optimal Parameters for Sequence Unlearning in Language Models","created":"2024-06-20","cited_by_count":10,"area":"safety"},{"id":"2501.19202","title":"Improving LLM Unlearning Robustness via Random Perturbations","created":"2025-01-31","cited_by_count":9,"area":"safety"},{"id":"2505.16723","title":"LLM Fingerprinting via Semantically Conditioned Watermarks","created":"2025-05-22","cited_by_count":9,"area":"safety"},{"id":"2410.13274","title":"Breaking Chains: Unraveling the Links in Multi-Hop Knowledge Unlearning","created":"2024-10-17","cited_by_count":8,"area":"safety"},{"id":"2509.25448","title":"Fingerprinting LLMs via Prompt Injection","created":"2025-09-29","cited_by_count":8,"area":"safety"},{"id":"2406.13551","title":"Mitigating Social Biases in Language Models through Unlearning","created":"2024-06-19","cited_by_count":7,"area":"safety"},{"id":"2509.20977","title":"CLUE: Conflict-guided Localization for LLM Unlearning Framework","created":"2025-09-25","cited_by_count":7,"area":"safety"},{"id":"2603.13461","title":"Purifying Generative LLMs from Backdoors without Prior Knowledge or Clean Reference","created":"2026-03-13","cited_by_count":7,"area":"safety"},{"id":"2607.19829","title":"DARWIN: Evolving Jailbreak Adversary and Guardrail for LLM Safety Evaluation and Protection","created":"2026-07-22","cited_by_count":6,"area":"safety"},{"id":"2407.20271","title":"Learn while Unlearn: An Iterative Unlearning Framework for Generative Language Models","created":"2024-07-25","cited_by_count":5,"area":"safety"},{"id":"2502.13996","title":"Beyond Single-Value Metrics: Evaluating and Enhancing LLM Unlearning with Cognitive Diagnosis","created":"2025-02-19","cited_by_count":5,"area":"safety"},{"id":"2502.15010","title":"Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language Models","created":"2025-02-20","cited_by_count":5,"area":"safety"},{"id":"2505.16567","title":"Watch your steps: Dormant Adversarial Behaviors that Activate upon LLM Finetuning","created":"2025-05-22","cited_by_count":5,"area":"safety"},{"id":"2509.22263","title":"Erase or Hide? Suppressing Spurious Unlearning Neurons for Robust Unlearning","created":"2025-09-26","cited_by_count":5,"area":"safety"},{"id":"2412.03107","title":"CredID: Credible Multi-Bit Watermark for Large Language Models Identification","created":"2024-12-04","cited_by_count":4,"area":"safety"},{"id":"2509.08000","title":"AntiDote: Bi-level Adversarial Training for Tamper-Resistant LLMs","created":"2025-09-06","cited_by_count":4,"area":"safety"},{"id":"2510.00761","title":"Downgrade to Upgrade: Optimizer Simplification Enhances Robustness in LLM Unlearning","created":"2025-10-01","cited_by_count":4,"area":"safety"},{"id":"2510.25117","title":"A Survey on Unlearning in Large Language Models","created":"2025-10-29","cited_by_count":4,"area":"safety"},{"id":"2603.23268","title":"SafeSeek: Universal Attribution of Safety Circuits in Language Models","created":"2026-03-24","cited_by_count":4,"area":"safety"},{"id":"2403.15740","title":"Protecting Copyrighted Material with Unique Identifiers in Large Language Model Training","created":"2024-03-23","cited_by_count":3,"area":"safety"},{"id":"2408.09459","title":"WPN: An Unlearning Method Based on N-pair Contrastive Learning in Language Models","created":"2024-08-18","cited_by_count":3,"area":"safety"},{"id":"2504.12108","title":"Entropy-Guided Watermarking for LLMs: A Test-Time Framework for Robust and Traceable Text Generation","created":"2025-04-16","cited_by_count":3,"area":"safety"},{"id":"2510.22376","title":"Label Smoothing Improves Gradient Ascent in LLM Unlearning","created":"2025-10-25","cited_by_count":3,"area":"safety"},{"id":"2511.07842","title":"Safety-Preserving PTQ via Contrastive Alignment Loss","created":"2025-11-11","cited_by_count":3,"area":"safety"},{"id":"2508.06467","title":"LLM Unlearning using Gradient Ratio-Based Influence Estimation and Noise Injection","created":"2025-08-08","cited_by_count":2,"area":"safety"},{"id":"2509.23362","title":"Dual-Space Smoothness for Robust and Balanced LLM Unlearning","created":"2025-09-27","cited_by_count":2,"area":"safety"},{"id":"2510.07775","title":"The Unintended Trade-off of AI Alignment:Balancing Hallucination Mitigation and Safety in LLMs","created":"2025-10-09","cited_by_count":2,"area":"safety"},{"id":"2606.19535","title":"FloatDoor: Platform-Triggered Backdoors in LLMs","created":"2026-06-17","cited_by_count":2,"area":"safety"},{"id":"2607.02513","title":"LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning","created":"2026-07-02","cited_by_count":2,"area":"safety"},{"id":"2402.03303","title":"Nevermind: Instruction Override and Moderation in Large Language Models","created":"2024-02-05","cited_by_count":1,"area":"safety"},{"id":"2406.05364","title":"Is On-Device AI Broken and Exploitable? Assessing the Trust and Ethics in Small Language Models","created":"2024-06-08","cited_by_count":1,"area":"safety"},{"id":"2504.06658","title":"A Neuro-inspired Interpretation of Unlearning in Large Language Models through Sample-level Unlearning Difficulty","created":"2025-04-09","cited_by_count":1,"area":"safety"},{"id":"2505.05619","title":"LiteLMGuard: Seamless and Lightweight On-Device Prompt Filtering for Safeguarding Small Language Models against Quantization-induced Risks and Vulnerabilities","created":"2025-05-08","cited_by_count":1,"area":"safety"},{"id":"2507.01752","title":"Tuning without Peeking: Provable Generalization Bounds and Robust LLM Post-Training","created":"2025-07-02","cited_by_count":1,"area":"safety"},{"id":"2507.08288","title":"Invariant-based Robust Weights Watermark for Large Language Models","created":"2025-07-11","cited_by_count":1,"area":"safety"},{"id":"2507.22160","title":"Strategic Deflection: Defending LLMs from Logit Manipulation","created":"2025-07-29","cited_by_count":1,"area":"safety"},{"id":"2509.05316","title":"Standard vs. Modular Sampling: Best Practices for Reliable LLM Unlearning","created":"2025-08-29","cited_by_count":1,"area":"safety"},{"id":"2511.19009","title":"Understanding and Mitigating Over-refusal for Large Language Models via Representation Intervention","created":"2025-11-24","cited_by_count":1,"area":"safety"},{"id":"2511.19566","title":"ModHiFi: Identifying High Fidelity predictive components for Model Modification","created":"2025-11-24","cited_by_count":1,"area":"safety"},{"id":"2601.21702","title":"Beyond Forgetting: Representation Misdirection Elicits Controllable Side Behaviors and Capabilities","created":"2026-01-29","cited_by_count":1,"area":"safety"},{"id":"2602.06623","title":"Do Prompts Guarantee Safety? Mitigating Toxicity from LLM Generations through Subspace Intervention","created":"2026-02-06","cited_by_count":1,"area":"safety"},{"id":"2606.03785","title":"Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs","created":"2026-06-02","cited_by_count":1,"area":"safety"},{"id":"2607.26849","title":"ToxScreen: Detecting Whether an LLM Has Been Poisoned","created":"2026-07-29","cited_by_count":1,"area":"safety"},{"id":"2505.17066","title":"Improving LLM Outputs Against Jailbreak Attacks with Expert Model Integration","created":"2025-05-18","cited_by_count":0,"area":"safety"},{"id":"2601.10566","title":"Representation-Aware Unlearning via Activation Signatures: From Suppression to Entity-Signature Erasure","created":"2026-01-15","cited_by_count":0,"area":"safety"},{"id":"2602.06630","title":"TrapSuffix: Proactive Defense Against Adversarial Suffixes in Jailbreaking","created":"2026-02-06","cited_by_count":0,"area":"safety"},{"id":"2603.18793","title":"Functional Subspace Watermarking for Large Language Models","created":"2026-03-19","cited_by_count":0,"area":"safety"},{"id":"2604.06154","title":"Exclusive Unlearning","created":"2026-04-07","cited_by_count":0,"area":"safety"},{"id":"2605.01699","title":"Probe-Geometry Alignment: Erasing the Cross-Sequence Memorization Signature Below Chance","created":"2026-05-03","cited_by_count":0,"area":"safety"},{"id":"2605.19147","title":"Be Kind, Rewrite: Benign Projections via Rewriting Defend Against LLM Data Poisoning Attacks","created":"2026-05-18","cited_by_count":0,"area":"safety"},{"id":"2606.17123","title":"LineageMark: Multi-user White-box Watermarking for Contribution Tracing in Model Derivation Chains","created":"2026-06-15","cited_by_count":0,"area":"safety"},{"id":"2607.20435","title":"Making Open-Source Text LLM Watermarks Durable Against Merging","created":"2026-05-11","cited_by_count":0,"area":"safety"},{"id":"2607.25375","title":"Inspect India Evals: An Open Benchmarking Framework for Evaluating Large Language Models in the Indian Linguistic and Cultural Context","created":"2026-07-28","cited_by_count":0,"area":"safety"},{"id":"2608.12652","title":"Excess Separability: Nuisance-Controlled Residual-Stream Probing for Benchmark Contamination Detection","created":"2026-08-12","cited_by_count":0,"area":"safety"},{"id":"2608.14229","title":"The More Popular, The Harder to Forget: Adaptive Popularity for LLM Unlearning","created":"2026-08-14","cited_by_count":0,"area":"safety"},{"id":"2608.20960","title":"Can Scientific Claims Be Removed from Large Language Models? A Systematic Evaluation of Claim-Level Unlearning","created":"2026-08-21","cited_by_count":0,"area":"safety"},{"id":"2608.27899","title":"OpenStamp: A Watermark for Open-Source Language Models","created":"2026-08-28","cited_by_count":0,"area":"safety"},{"id":"2609.00051","title":"From Detection to Refusal: Safer LLMs via Circuit-Guided Weight Scaling","created":"2026-08-30","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":"2609.29948","title":"ENDOPROMPT: Victim-Side Pseudo-References for Utility Degradation","created":"2026-09-24","cited_by_count":0,"area":"safety"},{"id":"2610.00348","title":"Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation","created":"2026-09-29","cited_by_count":0,"area":"safety"}],"next_cursor":"100"}