{"total":52,"rows":[{"id":"2502.17424","title":"Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs","created":"2025-02-24","cited_by_count":172,"area":"safety"},{"id":"2507.11878","title":"LLMs Encode Harmfulness and Refusal Separately","created":"2025-07-16","cited_by_count":88,"area":"safety"},{"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":"2508.00910","title":"Cyber-Zero: Training Cybersecurity Agents without Runtime","created":"2025-07-29","cited_by_count":22,"area":"agents"},{"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":"2508.03153","title":"Estimating Worst-Case Frontier Risks of Open-Weight LLMs","created":"2025-08-05","cited_by_count":19,"area":null},{"id":"2502.02260","title":"Position: Adversarial ML for LLMs Is Not Making Any Progress","created":"2025-02-04","cited_by_count":15,"area":null},{"id":"2508.18370","title":"Training Language Model Agents to Find Vulnerabilities with CTF-Dojo","created":"2025-08-25","cited_by_count":15,"area":"agents"},{"id":"2505.16559","title":"CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning","created":"2025-05-22","cited_by_count":11,"area":"safety"},{"id":"2508.07172","title":"Gradient Surgery for Safe LLM Fine-Tuning","created":"2025-08-10","cited_by_count":11,"area":"alignment"},{"id":"2505.12186","title":"Self-Destructive Language Model","created":"2025-05-18","cited_by_count":10,"area":"safety"},{"id":"2506.20702","title":"The Singapore Consensus on Global AI Safety Research Priorities","created":"2025-06-25","cited_by_count":10,"area":null},{"id":"2507.11630","title":"Jailbreak-Tuning: Models Efficiently Learn Jailbreak Susceptibility","created":"2025-07-15","cited_by_count":9,"area":"safety"},{"id":"2502.11910","title":"Adversarial Alignment for LLMs Requires Simpler, Reproducible, and More Measurable Objectives","created":"2025-02-17","cited_by_count":8,"area":null},{"id":"2505.18384","title":"Dynamic Risk Assessments for Offensive Cybersecurity Agents","created":"2025-05-23","cited_by_count":8,"area":"agents"},{"id":"2506.06414","title":"Benchmarking Misuse Mitigation Against Covert Adversaries","created":"2025-06-06","cited_by_count":8,"area":null},{"id":"2505.23270","title":"Does Machine Unlearning Truly Remove Knowledge?","created":"2025-05-29","cited_by_count":7,"area":"safety"},{"id":"2502.19537","title":"No, of Course I Can! Deeper Fine-Tuning Attacks That Bypass Token-Level Safety Mechanisms","created":"2025-02-26","cited_by_count":6,"area":"safety"},{"id":"2506.07452","title":"When Style Breaks Safety: Defending LLMs Against Superficial Style Alignment","created":"2025-06-09","cited_by_count":6,"area":"safety"},{"id":"2603.00498","title":"Antibody: Strengthening Defense Against Harmful Fine-Tuning for Large Language Models via Attenuating Harmful Gradient Influence","created":"2026-02-28","cited_by_count":6,"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":"2601.22313","title":"Hair-Trigger Alignment: Black-Box Evaluation Cannot Guarantee Post-Update Alignment","created":"2026-01-29","cited_by_count":4,"area":null},{"id":"2602.10568","title":"Gauss-Newton Unlearning for the LLM Era","created":"2026-02-11","cited_by_count":4,"area":"safety"},{"id":"2503.20848","title":"The Backfiring Effect of Weak AI Safety Regulation","created":"2025-03-26","cited_by_count":3,"area":null},{"id":"2512.01899","title":"Provably Safe Model Updates","created":"2025-12-01","cited_by_count":3,"area":null},{"id":"2604.09544","title":"Large Language Models Generate Harmful Responses Using a Distinct Mechanism, Shared Across Harm Types","created":"2026-04-10","cited_by_count":3,"area":"safety"},{"id":"2502.15836","title":"Soft Token Attacks Cannot Reliably Audit Unlearning in Large Language Models","created":"2025-02-20","cited_by_count":2,"area":"safety"},{"id":"2503.16833","title":"The Model Hears You: Audio Language Model Deployments Should Consider the Principle of Least Privilege","created":"2025-03-21","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":"2510.12200","title":"HackWorld: Evaluating Computer-Use Agents on Exploiting Web Application Vulnerabilities","created":"2025-10-14","cited_by_count":2,"area":"agents"},{"id":"2602.06911","title":"TamperBench: Systematically Stress-Testing LLM Safety Under Fine-Tuning and Tampering","created":"2026-02-06","cited_by_count":2,"area":"safety"},{"id":"2605.14605","title":"A Few Steps Further: Why Defenses Against Malicious Finetuning Erode Under Continued Training","created":"2026-05-14","cited_by_count":2,"area":null},{"id":"2502.15720","title":"Training AI to be Loyal","created":"2025-01-27","cited_by_count":1,"area":null},{"id":"2510.27629","title":"Best Practices for Biorisk Evaluations on Open-Weight Bio-Foundation Models","created":"2025-10-31","cited_by_count":1,"area":null},{"id":"2601.01584","title":"Steerability of Instrumental-Convergence Tendencies in LLMs","created":"2026-01-04","cited_by_count":1,"area":null},{"id":"2605.25073","title":"Security in the Fine-Tuning Lifecycle of Large Language Models: Threats, Defenses,Evaluation, and Future Directions","created":"2026-05-24","cited_by_count":1,"area":"safety"},{"id":"2605.26526","title":"Open-Weight LLM Fine-Tuning Defenses are Susceptible to Simple Attacks","created":"2026-05-26","cited_by_count":1,"area":"safety"},{"id":"2606.25059","title":"What Does It Mean to Break a Distillation Defense?","created":"2026-06-23","cited_by_count":1,"area":null},{"id":"2607.12406","title":"Isolation as a First-Class Principle for LLM-Agent System Safety: Concepts, Taxonomy, Challenges and Future Directions","created":"2026-07-14","cited_by_count":1,"area":"agents"},{"id":"2607.13039","title":"Safeguard-Conditioned Uplift: Measuring Utility-Risk Frontiers for Dual-Use Biology Assistants","created":"2026-06-12","cited_by_count":1,"area":null},{"id":"2607.22929","title":"Distribution-Specific Curvature Control with Finite-Sample Guarantees for Open-Weight Safety","created":"2026-07-24","cited_by_count":1,"area":null},{"id":"2507.10106","title":"BlueGlass: A Framework for Composite AI Safety","created":"2025-07-14","cited_by_count":0,"area":null},{"id":"2602.17418","title":"A Privacy by Design Framework for Large Language Model-Based Applications for Children","created":"2026-02-19","cited_by_count":0,"area":null},{"id":"2605.10777","title":"Locking Pretrained Weights via Deep Low-Rank Residual Distillation","created":"2026-05-11","cited_by_count":0,"area":null},{"id":"2606.07970","title":"Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks","created":"2026-06-06","cited_by_count":0,"area":"alignment"},{"id":"2608.05045","title":"Gradient Immunity: Null-Space Resistance to Malicious Fine-Tuning","created":"2026-08-05","cited_by_count":0,"area":"alignment"},{"id":"2608.17202","title":"Fool's Gold: Defensive Deception Against Safety-Removal Attacks on Open-Weight Models","created":"2026-08-17","cited_by_count":0,"area":"alignment"},{"id":"2609.00519","title":"The Safeguard Worked. Is the LLM System Safer?","created":"2026-09-01","cited_by_count":0,"area":null},{"id":"2609.20027","title":"Can Data Attribution Filter Out Subliminal Learning? Not Reliably","created":"2026-09-17","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":"2610.01323","title":"TRACE: Trajectory Return Attribution and Contrastive Erasure for Multi-Turn Safety","created":"2026-10-01","cited_by_count":0,"area":"alignment"},{"id":"2610.03124","title":"The Fragility of Trigger-Tag Mechanisms for Misuse Detection in Open-Weight LLMs","created":"2026-10-02","cited_by_count":0,"area":"safety"}],"next_cursor":null}