{"total":81,"rows":[{"id":"2409.01586","title":"Booster: Tackling Harmful Fine-tuning for Large Language Models via Attenuating Harmful Perturbation","created":"2024-09-03","cited_by_count":92,"area":null},{"id":"2407.09121","title":"Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training","created":"2024-07-12","cited_by_count":69,"area":null},{"id":"2510.26243","title":"Angular Steering: Behavior Control via Rotation in Activation Space","created":"2025-10-30","cited_by_count":53,"area":null},{"id":"2412.07097","title":"On Evaluating the Durability of Safeguards for Open-Weight LLMs","created":"2024-12-10","cited_by_count":52,"area":null},{"id":"2406.18664","title":"Evaluating Copyright Takedown Methods for Language Models","created":"2024-06-26","cited_by_count":43,"area":null},{"id":"2504.16980","title":"Safety Pretraining: Toward the Next Generation of Safe AI","created":"2025-04-23","cited_by_count":43,"area":null},{"id":"2502.05209","title":"Model Tampering Attacks Enable More Rigorous Evaluations of LLM Capabilities","created":"2025-02-03","cited_by_count":39,"area":null},{"id":"2504.02922","title":"Overcoming Sparsity Artifacts in Crosscoders to Interpret Chat-Tuning","created":"2025-04-03","cited_by_count":34,"area":null},{"id":"2407.18369","title":"AI Safety in Generative AI Large Language Models: A Survey","created":"2024-07-06","cited_by_count":33,"area":null},{"id":"2501.18100","title":"Panacea: Mitigating Harmful Fine-tuning for Large Language Models via Post-fine-tuning Perturbation","created":"2025-01-30","cited_by_count":20,"area":null},{"id":"2506.08473","title":"AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin","created":"2025-06-10","cited_by_count":19,"area":null},{"id":"2509.21651","title":"Can AI Perceive Physical Danger and Intervene?","created":"2025-09-25","cited_by_count":14,"area":null},{"id":"2505.17646","title":"Unveiling the Basin-Like Loss Landscape in Large Language Models","created":"2025-05-23","cited_by_count":13,"area":null},{"id":"2506.01317","title":"T-SHIRT: Token-Selective Hierarchical Data Selection for Instruction Tuning","created":"2025-06-02","cited_by_count":13,"area":null},{"id":"2510.05571","title":"Presenting a Paper is an Art: Self-Improvement Aesthetic Agents for Academic Presentations","created":"2025-10-07","cited_by_count":13,"area":null},{"id":"2502.14828","title":"Fundamental Limitations in Pointwise Defences of LLM Finetuning APIs","created":"2025-02-20","cited_by_count":10,"area":null},{"id":"2507.10540","title":"FusionFactory: Fusing LLM Capabilities with Multi-LLM Log Data","created":"2025-07-14","cited_by_count":10,"area":null},{"id":"2505.17441","title":"Discovering Forbidden Topics in Language Models","created":"2025-05-23","cited_by_count":9,"area":null},{"id":"2508.20697","title":"Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning","created":"2025-08-28","cited_by_count":8,"area":null},{"id":"2509.20393","title":"The Secret Agenda: LLMs Strategically Lie and Our Current Safety Tools Are Blind","created":"2025-09-23","cited_by_count":8,"area":null},{"id":"2506.19882","title":"Position: Machine Learning Conferences Should Establish a \"Refutations and Critiques\" Track","created":"2025-06-24","cited_by_count":7,"area":null},{"id":"2509.22367","title":"What Is The Political Content in LLMs' Pre- and Post-Training Data?","created":"2025-09-26","cited_by_count":7,"area":null},{"id":"2602.05910","title":"Chunky Post-Training: Data Driven Failures of Generalization","created":"2026-02-05","cited_by_count":7,"area":null},{"id":"2409.03274","title":"Recent Advances in Attack and Defense Approaches of Large Language Models","created":"2024-09-05","cited_by_count":6,"area":null},{"id":"2509.21155","title":"Learning the Wrong Lessons: Syntactic-Domain Spurious Correlations in Language Models","created":"2025-09-25","cited_by_count":6,"area":null},{"id":"2510.00857","title":"ManagerBench: Evaluating the Safety-Pragmatism Trade-off in Autonomous LLMs","created":"2025-10-01","cited_by_count":6,"area":null},{"id":"2603.03308","title":"Old Habits Die Hard: How Conversational History Geometrically Traps LLMs","created":"2026-02-08","cited_by_count":6,"area":null},{"id":"2505.23724","title":"SC-LoRA: Balancing Efficient Fine-tuning and Knowledge Preservation via Subspace-Constrained LoRA","created":"2025-05-29","cited_by_count":5,"area":null},{"id":"2511.00382","title":"Efficiency vs. Alignment: Investigating Safety and Fairness Risks in Parameter-Efficient Fine-Tuning of LLMs","created":"2025-11-01","cited_by_count":4,"area":null},{"id":"2607.26654","title":"Constitutional Midtraining: Content Presence Drives Alignment Gains","created":"2026-07-29","cited_by_count":4,"area":null},{"id":"2508.11281","title":"ToxiFrench: Benchmarking and Enhancing Language Models via CoT Fine-Tuning for French Toxicity Detection","created":"2025-08-15","cited_by_count":3,"area":null},{"id":"2511.13052","title":"Learning from the Undesirable: Robust Adaptation of Language Models without Forgetting","created":"2025-11-17","cited_by_count":3,"area":null},{"id":"2512.23779","title":"Prompt-Induced Over-Generation as Denial-of-Service: A Black-Box Attack-Side Benchmark","created":"2025-12-29","cited_by_count":3,"area":null},{"id":"2602.16729","title":"Intent Laundering: AI Safety Datasets Are Not What They Seem","created":"2026-02-17","cited_by_count":3,"area":null},{"id":"2506.05970","title":"Let's Put Ourselves in Sally's Shoes: Shoes-of-Others Prefilling Improves Theory of Mind in Large Language Models","created":"2025-06-06","cited_by_count":2,"area":null},{"id":"2507.11544","title":"The Safety Gap Toolkit: Evaluating Hidden Dangers of Open-Source Models","created":"2025-07-08","cited_by_count":2,"area":null},{"id":"2509.21456","title":"Diagnosing the Performance Trade-off in Moral Alignment: A Case Study on Gender Stereotypes","created":"2025-09-25","cited_by_count":2,"area":null},{"id":"2602.21215","title":"Inference-time Alignment via Sparse Junction Steering","created":"2026-01-30","cited_by_count":2,"area":null},{"id":"2606.13944","title":"LLMs Contain Multitudes: How Deployment Context Reshapes Model-Level Preferences and Values","created":"2026-06-11","cited_by_count":2,"area":null},{"id":"2607.09053","title":"An Emergent Mirage: Is Emergent Misalignment and Realignment Indeed a Robust Phenomenon?","created":"2026-07-10","cited_by_count":2,"area":null},{"id":"2608.28327","title":"Layered LLM Defenses as an Ensemble: Access Tiers, Inference Cost, and the Measured Failure Correlation Between Defense Layers","created":"2026-08-28","cited_by_count":2,"area":null},{"id":"2506.00047","title":"Risks of AI-driven product development and strategies for their mitigation","created":"2025-05-28","cited_by_count":1,"area":null},{"id":"2508.08855","title":"BiasGym: A Simple and Generalizable Framework for Analyzing and Removing Biases through Injection","created":"2025-08-12","cited_by_count":1,"area":null},{"id":"2603.00078","title":"Alignment Is Not Enough: A Relational Framework for Moral Standing in Human-AI Interaction","created":"2026-02-13","cited_by_count":1,"area":null},{"id":"2604.09688","title":"Immunizing 3D Gaussian Generative Models Against Unauthorized Fine-Tuning via Attribute-Space Traps","created":"2026-04-06","cited_by_count":1,"area":null},{"id":"2604.26577","title":"Benchmarking the Safety of Large Language Models for Robotic Health Attendant Control","created":"2026-04-29","cited_by_count":1,"area":null},{"id":"2606.00160","title":"DataShield: Safety-degrading Data Filtering for LLM Benign Instruction Fine-Tuning","created":"2026-05-29","cited_by_count":1,"area":null},{"id":"2606.10154","title":"Quality Is Not a Safety Proxy Under Quantization","created":"2026-06-08","cited_by_count":1,"area":null},{"id":"2606.18656","title":"The Wrong Kind of Right: Quantifying and Localizing Misfired Alignment in LLMs","created":"2026-06-17","cited_by_count":1,"area":null},{"id":"2506.05171","title":"Towards provable probabilistic safety for scalable embodied AI systems","created":"2025-06-05","cited_by_count":0,"area":null},{"id":"2507.13629","title":"Large Language Models in Cybersecurity: Applications, Vulnerabilities, and Defense Techniques","created":"2025-07-18","cited_by_count":0,"area":null},{"id":"2508.09994","title":"Whisper Smarter, not Harder: Adversarial Attack on Partial Suppression","created":"2025-07-30","cited_by_count":0,"area":null},{"id":"2512.10234","title":"InFerActive: Interactive Tree-Based Exploration of LLM Sampling for Safety Evaluation","created":"2025-12-11","cited_by_count":0,"area":null},{"id":"2601.03079","title":"Learning to Diagnose and Correct Moral Errors: Beyond Shallow Heuristics in Moral Alignment","created":"2026-01-06","cited_by_count":0,"area":null},{"id":"2601.18998","title":"Malicious Repurposing of Open Science Artefacts by Using Large Language Models","created":"2026-01-26","cited_by_count":0,"area":null},{"id":"2602.02258","title":"Alignment-Aware Model Adaptation via Feedback-Guided Optimization","created":"2026-02-02","cited_by_count":0,"area":null},{"id":"2602.17546","title":"Learning to Stay Safe: Adaptive Regularization Against Safety Degradation during Fine-Tuning","created":"2026-02-19","cited_by_count":0,"area":null},{"id":"2603.21155","title":"Can LLMs Fool Graph Learning? Exploring Universal Adversarial Attacks on Text-Attributed Graphs","created":"2026-03-22","cited_by_count":0,"area":null},{"id":"2604.07709","title":"IatroBench: A Pre-Registered Benchmark of Clinical Omission in Language Models","created":"2026-04-09","cited_by_count":0,"area":null},{"id":"2605.14194","title":"GradShield: Alignment Preserving Finetuning","created":"2026-05-13","cited_by_count":0,"area":null},{"id":"2605.24583","title":"Measuring Alignment-Induced Activation Shifts Correctly: A Template-Controlled Difference-in-Differences Protocol","created":"2026-05-23","cited_by_count":0,"area":null},{"id":"2606.16914","title":"Greed Is Learned: Visible Incentives as Reward-Hacking Triggers","created":"2026-06-15","cited_by_count":0,"area":null},{"id":"2606.23671","title":"Can LLMs Reliably Self-Report Adversarial Prefills, and How?","created":"2026-06-22","cited_by_count":0,"area":null},{"id":"2606.26515","title":"Forget, Anticipate and Adapt: Test Time Training for Long Videos","created":"2026-06-25","cited_by_count":0,"area":null},{"id":"2606.29441","title":"Closing the Activation-Cone Blind Spot: Response-Time Probing and Unified Defense","created":"2026-06-28","cited_by_count":0,"area":null},{"id":"2606.31876","title":"Harnessing Textual Refusal Directions for Multimodal Safety","created":"2026-06-30","cited_by_count":0,"area":null},{"id":"2607.04510","title":"Transplanting, inverting, and preventing a misalignment persona: method-conditional emergent misalignment in Qwen2.5","created":"2026-07-05","cited_by_count":0,"area":null},{"id":"2607.04846","title":"Pretraining Curricula Enable Selective Fine-tuning","created":"2026-07-06","cited_by_count":0,"area":null},{"id":"2607.20791","title":"Refusal-Gated Decoding: Preserving Refusal Behavior Under High-Temperature Sampling","created":"2026-07-22","cited_by_count":0,"area":null},{"id":"2608.00042","title":"Trustworthiness Costs of Domain Adaptation in Small Language Models:A Cross-Architecture Empirical Study","created":"2026-07-23","cited_by_count":0,"area":null},{"id":"2608.04322","title":"DataRx: Missingness-Aware Sampling for Safer Large Language Model Task-Specific Fine-Tuning","created":"2026-08-05","cited_by_count":0,"area":null},{"id":"2608.05158","title":"Safe Evolution with Circuit Anchors","created":"2026-05-24","cited_by_count":0,"area":null},{"id":"2609.02707","title":"Door-in-the-Face Requests and Refusal Behaviour in Large Language Models","created":"2026-09-02","cited_by_count":0,"area":null},{"id":"2609.22153","title":"SafeTune: A Unified Faithful Library for Auditing and Repairing Safety Drift in Fine-Tuned LLMs","created":"2026-08-26","cited_by_count":0,"area":null},{"id":"2609.33150","title":"Generalization Dynamics of LM Pre-training","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.33559","title":"Fine Until Fine-Tuned: Repeated Solutions Make Reasoning Fragile","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.38205","title":"The System Prompt Illusion: How Instruction Preambles Modify Computation in Language Models","created":"2026-09-23","cited_by_count":0,"area":null},{"id":"2609.39788","title":"Safety of Latent Communication in Multi-Agent Systems","created":"2026-09-30","cited_by_count":0,"area":null},{"id":"2610.04316","title":"From Latent Space to Jacobian Space: Measuring, Evading, and Training Against Safety-Content Accessibility","created":"2026-10-03","cited_by_count":0,"area":null},{"id":"2610.16043","title":"Beyond the greenwash: do LLMs' ecologically sound decisions hold up under pressure?","created":"2026-10-12","cited_by_count":0,"area":null},{"id":"2610.17728","title":"Safety for Whom? Benchmarking Understanding Beyond Refusal in Egocentric Multimodal Assistants","created":"2026-10-13","cited_by_count":0,"area":null}],"next_cursor":null}