{"total":72,"rows":[{"id":"2512.15688","title":"BashArena: A Control Setting for Highly Privileged AI Agents","created":"2025-12-17","cited_by_count":17,"area":"agents"},{"id":"2603.07084","title":"Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR","created":"2026-03-07","cited_by_count":17,"area":"alignment"},{"id":"2602.03085","title":"The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers","created":"2026-02-03","cited_by_count":14,"area":"safety"},{"id":"2512.05648","title":"Beyond Data Filtering: Knowledge Localization for Capability Removal in LLMs","created":"2025-12-05","cited_by_count":11,"area":"safety"},{"id":"2510.25771","title":"Gaperon: A Peppered English-French Generative Language Model Suite","created":"2025-10-29","cited_by_count":9,"area":null},{"id":"2602.06258","title":"GRP-Obliteration: Unaligning LLMs With a Single Unlabeled Prompt","created":"2026-02-05","cited_by_count":7,"area":"alignment"},{"id":"2605.23645","title":"Learning Through Noise: Why Subliminal Learning Works and When It Fails","created":"2026-05-22","cited_by_count":7,"area":null},{"id":"2601.15220","title":"Privacy Collapse: Benign Fine-Tuning Can Break Contextual Privacy in Language Models","created":"2026-01-21","cited_by_count":6,"area":null},{"id":"2602.22303","title":"Training Agents to Self-Report Misbehavior","created":"2026-02-25","cited_by_count":6,"area":"agents"},{"id":"2602.09987","title":"Infusion: Shaping Model Behavior by Editing Training Data via Influence Functions","created":"2026-02-10","cited_by_count":5,"area":null},{"id":"2602.10568","title":"Gauss-Newton Unlearning for the LLM Era","created":"2026-02-11","cited_by_count":4,"area":"safety"},{"id":"2602.18700","title":"Watermarking LLM Agent Trajectories","created":"2026-02-21","cited_by_count":4,"area":"agents"},{"id":"2604.13076","title":"Alignment midtraining for animals","created":"2026-03-21","cited_by_count":4,"area":null},{"id":"2606.19370","title":"Human-like autonomy emerges from self-play and a pinch of human data","created":"2026-06-11","cited_by_count":4,"area":"driving"},{"id":"2511.11020","title":"Data Poisoning Vulnerabilities Across Healthcare AI Architectures: A Security Threat Analysis","created":"2025-11-14","cited_by_count":3,"area":"safety"},{"id":"2602.00169","title":"Towards Agentic Intelligence for Materials Science","created":"2026-01-29","cited_by_count":3,"area":"agents"},{"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":"2603.02960","title":"Architecting Trust in Artificial Epistemic Agents","created":"2026-03-03","cited_by_count":3,"area":"agents"},{"id":"2603.03371","title":"Sleeper Cell: Injecting Latent Malice Temporal Backdoors into Tool-Using LLMs","created":"2026-03-02","cited_by_count":3,"area":"agents"},{"id":"2604.16423","title":"Shifting the Gradient: Understanding How Defensive Training Methods Protect Language Model Integrity","created":"2026-04-03","cited_by_count":3,"area":null},{"id":"2602.10382","title":"Language Triggers Hijack Language Circuits: A Mechanistic Analysis of Backdoor Behaviors in Large Language Models","created":"2026-02-11","cited_by_count":2,"area":"safety"},{"id":"2604.23593","title":"When AI reviews science: Can we trust the referee?","created":"2026-04-26","cited_by_count":2,"area":null},{"id":"2607.26339","title":"RAGuard: A Layered Defense Framework for Retrieval-Augmented Generation Systems Against Data Poisoning","created":"2026-07-28","cited_by_count":2,"area":"rag"},{"id":"2510.26829","title":"Layer of Truth: Probing Belief Shifts under Continual Pre-Training Poisoning","created":"2025-10-29","cited_by_count":1,"area":null},{"id":"2602.03899","title":"Byzantine Machine Learning: MultiKrum and an optimal notion of robustness","created":"2026-02-03","cited_by_count":1,"area":null},{"id":"2602.13427","title":"Backdooring Bias in Large Language Models","created":"2026-02-13","cited_by_count":1,"area":"safety"},{"id":"2602.17729","title":"Stop Saying \"AI\"","created":"2026-02-18","cited_by_count":1,"area":null},{"id":"2603.02262","title":"Silent Sabotage During Fine-Tuning: Few-Shot Rationale Poisoning of Compact Medical LLMs","created":"2026-02-28","cited_by_count":1,"area":"safety"},{"id":"2603.10806","title":"Backdoor Directions in Vision Transformers","created":"2026-03-11","cited_by_count":1,"area":"safety"},{"id":"2605.06846","title":"Narrow Secret Loyalty Dodges Black-Box Audits","created":"2026-05-07","cited_by_count":1,"area":"safety"},{"id":"2605.10862","title":"RUBEN: Rule-Based Explanations for Retrieval-Augmented LLM Systems","created":"2026-05-11","cited_by_count":1,"area":"rag"},{"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":"2607.15267","title":"Pretraining Data Can Be Poisoned through Computational Propaganda","created":"2026-07-16","cited_by_count":1,"area":null},{"id":"2607.23394","title":"Inference-Time Consensus for Mitigating Hidden Behaviors from LLM Fine-Tuning","created":"2026-07-25","cited_by_count":1,"area":null},{"id":"2607.26849","title":"ToxScreen: Detecting Whether an LLM Has Been Poisoned","created":"2026-07-29","cited_by_count":1,"area":"safety"},{"id":"2608.08131","title":"Compositional Threat Analysis of Latent Compromise in LLM Agent Systems: The Order 66 Scenario","created":"2026-08-08","cited_by_count":1,"area":"agents"},{"id":"2609.17380","title":"OPEN-1B: A Fully Auditable Training Run","created":"2026-09-15","cited_by_count":1,"area":null},{"id":"2510.22689","title":"Rule-Based Explanations for Retrieval-Augmented LLM Systems","created":"2025-10-26","cited_by_count":0,"area":"rag"},{"id":"2511.02600","title":"On The Dangers of Poisoned LLMs In Security Automation","created":"2025-11-04","cited_by_count":0,"area":null},{"id":"2511.13753","title":"Robustness of LLM-enabled vehicle trajectory prediction under data security threats","created":"2025-11-14","cited_by_count":0,"area":null},{"id":"2512.02043","title":"Mirror, Mirror on the Wall -- Which is the Best Model of Them All?","created":"2025-11-25","cited_by_count":0,"area":null},{"id":"2512.02786","title":"FiMMIA: scaling semantic perturbation-based membership inference across modalities","created":"2025-12-02","cited_by_count":0,"area":"safety"},{"id":"2603.07401","title":"VIVECaption: A Split Approach to Caption Quality Improvement","created":"2026-03-08","cited_by_count":0,"area":null},{"id":"2603.17174","title":"Detecting Data Poisoning in Code Generation LLMs via Black-Box, Vulnerability-Oriented Scanning","created":"2026-03-17","cited_by_count":0,"area":"safety"},{"id":"2604.12548","title":"DeepSeek Robustness Against Semantic-Character Dual-Space Mutated Prompt Injection","created":"2026-04-14","cited_by_count":0,"area":"safety"},{"id":"2604.22191","title":"Behavioral Canaries: Auditing Private Retrieved Context Usage in RL Fine-Tuning","created":"2026-04-24","cited_by_count":0,"area":"agents"},{"id":"2604.23238","title":"Hiding in Plain Sight: Detectability-Aware Antidistillation of Reasoning Models","created":"2026-04-25","cited_by_count":0,"area":"lr"},{"id":"2604.25639","title":"Large language models eroding science understanding: an empirical study of malignment","created":"2026-04-28","cited_by_count":0,"area":null},{"id":"2605.16325","title":"Phase Transitions in Driven Informational Systems: A Two-Field Perspective on Learning Theory and Non-Equilibrium Chemistry","created":"2026-05-05","cited_by_count":0,"area":null},{"id":"2605.22481","title":"When Stronger Triggers Backfire: A High-Dimensional Theory of Backdoor Attacks","created":"2026-05-21","cited_by_count":0,"area":null},{"id":"2606.05919","title":"Finding Most Influential Sets","created":"2026-06-04","cited_by_count":0,"area":null},{"id":"2606.07245","title":"AI Sovereignty: A Qualitative Model of Strategic Competition as AI Becomes an Instrument of National Power","created":"2026-06-05","cited_by_count":0,"area":null},{"id":"2606.14816","title":"A Security Analysis of Long-Horizon Agentic AI Systems: Threats, Evaluation, and Framework Development","created":"2026-06-12","cited_by_count":0,"area":"agents"},{"id":"2606.16242","title":"Rapid Poison: Practical Poisoning Attacks Against the Rapid Response Framework","created":"2026-06-15","cited_by_count":0,"area":"safety"},{"id":"2607.01729","title":"DRL-CLBA: A Clean Label Backdoor Attack for Speech Classification via DDPG Reinforcement Learning","created":"2026-07-02","cited_by_count":0,"area":"safety"},{"id":"2607.04146","title":"!Imperio, smolVLA: The Implications of Data Poisoning on Open Source Robotics","created":"2026-07-05","cited_by_count":0,"area":"safety"},{"id":"2607.06643","title":"The Power of Backdoor Absorption in Community Training","created":"2026-07-07","cited_by_count":0,"area":null},{"id":"2607.22657","title":"Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS","created":"2026-06-27","cited_by_count":0,"area":"safety"},{"id":"2608.14667","title":"Position: AI Agents in Scientific Teams Should Be Studied as Human-Agent Systems","created":"2026-08-02","cited_by_count":0,"area":"agents"},{"id":"2608.27954","title":"Not to Break, but to Attest: Adversarial Probes for Privacy-Preserving LLM Verification","created":"2026-08-28","cited_by_count":0,"area":null},{"id":"2609.10883","title":"Story Imprinting: AI Assistants Absorb Traits from Human Characters They Resemble","created":"2026-09-09","cited_by_count":0,"area":null},{"id":"2609.16694","title":"Toward Secure AI-Powered Penetration Testing Agents: Security Threats, Guardrails, and Architectural Perspectives","created":"2026-09-15","cited_by_count":0,"area":"agents"},{"id":"2609.18560","title":"The evolution of sex for artificial intelligence: a population-genetic framework for multigenerational model populations","created":"2026-09-16","cited_by_count":0,"area":null},{"id":"2609.32021","title":"SilentCall: Hidden Tool-Call Backdoors in Open-Weight Agents, and How to Catch Them","created":"2026-09-25","cited_by_count":0,"area":"agents"},{"id":"2609.32288","title":"A Solvable Theory of Pre-training Data Poisoning: Regime-Dependent Scaling Exponents","created":"2026-09-26","cited_by_count":0,"area":"safety"},{"id":"2609.33915","title":"Training Witnesses: Trusting the Training without Trusting the Trainer","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.33985","title":"The Privacy Fallacy of Crowdsourced Fine-Tuning: Extracting Proprietary Data via Topic-Based Poisoning","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.36117","title":"Why Backdooring Neural Networks is so Easy?","created":"2026-09-28","cited_by_count":0,"area":null},{"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"},{"id":"2610.02886","title":"Misinformation Without Triggers: From Factual Answers to Downstream Decisions","created":"2026-10-02","cited_by_count":0,"area":null},{"id":"2610.06994","title":"TARE: Weigh a Never-Poisoned Twin Before Reading Backdoor-Defense Costs","created":"2026-10-04","cited_by_count":0,"area":null},{"id":"2610.14487","title":"AlignPoison: Conditional Targeted Poisoning of Generative Recommenders via Decoding-Aligned Prefixes","created":"2026-10-10","cited_by_count":0,"area":null}],"next_cursor":null}