{"total":90,"rows":[{"id":"2306.17439","title":"Provable Robust Watermarking for AI-Generated Text","created":"2023-06-30","cited_by_count":299,"area":"safety"},{"id":"2407.15549","title":"Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs","created":"2024-07-22","cited_by_count":173,"area":"safety"},{"id":"2502.17424","title":"Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs","created":"2025-02-24","cited_by_count":172,"area":"safety"},{"id":"2306.17194","title":"On the Exploitability of Instruction Tuning","created":"2023-06-28","cited_by_count":116,"area":"safety"},{"id":"2403.05030","title":"Defending Against Unforeseen Failure Modes with Latent Adversarial Training","created":"2024-03-08","cited_by_count":88,"area":"safety"},{"id":"2310.13828","title":"Nightshade: Prompt-Specific Poisoning Attacks on Text-to-Image Generative Models","created":"2023-10-20","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":"2303.03323","title":"CleanCLIP: Mitigating Data Poisoning Attacks in Multimodal Contrastive Learning","created":"2023-03-06","cited_by_count":63,"area":"safety"},{"id":"2407.14937","title":"Operationalizing a Threat Model for Red-Teaming Large Language Models (LLMs)","created":"2024-07-20","cited_by_count":46,"area":"safety"},{"id":"2410.08811","title":"PoisonBench: Assessing Large Language Model Vulnerability to Data Poisoning","created":"2024-10-11","cited_by_count":33,"area":"safety"},{"id":"2411.12768","title":"CROW: Eliminating Backdoors from Large Language Models via Internal Consistency Regularization","created":"2024-11-18","cited_by_count":33,"area":"safety"},{"id":"2411.17693","title":"Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats","created":"2024-11-26","cited_by_count":31,"area":"safety"},{"id":"2402.14015","title":"Corrective Machine Unlearning","created":"2024-02-21","cited_by_count":29,"area":"safety"},{"id":"2302.12461","title":"Analyzing And Editing Inner Mechanisms Of Backdoored Language Models","created":"2023-02-24","cited_by_count":23,"area":"safety"},{"id":"2402.02160","title":"Data Poisoning for In-context Learning","created":"2024-02-03","cited_by_count":21,"area":"safety"},{"id":"2503.22759","title":"Data Poisoning in Deep Learning: A Survey","created":"2025-03-27","cited_by_count":20,"area":"safety"},{"id":"2310.05862","title":"Better Safe than Sorry: Pre-training CLIP against Targeted Data Poisoning and Backdoor Attacks","created":"2023-10-05","cited_by_count":19,"area":"safety"},{"id":"2406.05670","title":"Certified Robustness to Data Poisoning in Gradient-Based Training","created":"2024-06-09","cited_by_count":18,"area":"safety"},{"id":"2503.00596","title":"BadJudge: Backdoor Vulnerabilities of LLM-as-a-Judge","created":"2025-03-01","cited_by_count":18,"area":"safety"},{"id":"2409.15398","title":"Attack Atlas: A Practitioner's Perspective on Challenges and Pitfalls in Red Teaming GenAI","created":"2024-09-23","cited_by_count":16,"area":"safety"},{"id":"2601.07072","title":"Overcoming the Retrieval Barrier: Indirect Prompt Injection in the Wild for LLM Systems","created":"2026-01-11","cited_by_count":16,"area":"safety"},{"id":"2407.04108","title":"Future Events as Backdoor Triggers: Investigating Temporal Vulnerabilities in LLMs","created":"2024-07-04","cited_by_count":15,"area":"safety"},{"id":"2409.19993","title":"Mitigating Backdoor Threats to Large Language Models: Advancement and Challenges","created":"2024-09-30","cited_by_count":15,"area":"safety"},{"id":"2307.10163","title":"Rethinking Backdoor Attacks","created":"2023-07-19","cited_by_count":14,"area":"safety"},{"id":"2310.18603","title":"Large Language Models Are Better Adversaries: Exploring Generative Clean-Label Backdoor Attacks Against Text Classifiers","created":"2023-10-28","cited_by_count":14,"area":"safety"},{"id":"2405.07667","title":"Simulate and Eliminate: Revoke Backdoors for Generative Large Language Models","created":"2024-05-13","cited_by_count":12,"area":"safety"},{"id":"2412.20392","title":"Defending Multimodal Backdoored Models by Repulsive Visual Prompt Tuning","created":"2024-12-29","cited_by_count":12,"area":"safety"},{"id":"2409.15968","title":"Adversarial Backdoor Defense in CLIP","created":"2024-09-24","cited_by_count":10,"area":"safety"},{"id":"2406.11007","title":"Threat Modelling and Risk Analysis for Large Language Model (LLM)-Powered Applications","created":"2024-06-16","cited_by_count":8,"area":"safety"},{"id":"2502.19567","title":"Atlas: A Framework for ML Lifecycle Provenance & Transparency","created":"2025-02-26","cited_by_count":8,"area":"safety"},{"id":"2505.17730","title":"Redirection for Erasing Memory (REM): Towards a universal unlearning method for corrupted data","created":"2025-05-23","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":"2410.02916","title":"LLM Safeguard is a Double-Edged Sword: Exploiting False Positives for Denial-of-Service Attacks","created":"2024-10-03","cited_by_count":6,"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":"2510.17021","title":"Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearning","created":"2025-10-19","cited_by_count":6,"area":"safety"},{"id":"2405.15269","title":"Test-Time Multimodal Backdoor Detection by Contrastive Prompting","created":"2024-05-24","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":"2601.02680","title":"Adversarial Contrastive Learning for LLM Quantization Attacks","created":"2026-01-06","cited_by_count":5,"area":"safety"},{"id":"2602.15671","title":"Revisiting Backdoor Threat in Federated Instruction Tuning from a Signal Aggregation Perspective","created":"2026-02-17","cited_by_count":5,"area":"safety"},{"id":"2410.21750","title":"Learning and Unlearning of Fabricated Knowledge in Language Models","created":"2024-10-29","cited_by_count":4,"area":"safety"},{"id":"2602.10568","title":"Gauss-Newton Unlearning for the LLM Era","created":"2026-02-11","cited_by_count":4,"area":"safety"},{"id":"2602.16944","title":"Exact Certification of Data-Poisoning Attacks Using Mixed-Integer Programming","created":"2026-02-18","cited_by_count":4,"area":"safety"},{"id":"2404.08631","title":"FCert: Certifiably Robust Few-Shot Classification in the Era of Foundation Models","created":"2024-04-12","cited_by_count":3,"area":"safety"},{"id":"2412.00761","title":"Learning to Forget using Hypernetworks","created":"2024-12-01","cited_by_count":3,"area":"safety"},{"id":"2412.02113","title":"Trust & Safety of LLMs and LLMs in Trust & Safety","created":"2024-12-03","cited_by_count":3,"area":"safety"},{"id":"2503.07697","title":"PoisonedParrot: Subtle Data Poisoning Attacks to Elicit Copyright-Infringing Content from Large Language Models","created":"2025-03-10","cited_by_count":3,"area":"safety"},{"id":"2509.18717","title":"Pre-training CLIP against Data Poisoning with Optimal Transport-based Matching and Alignment","created":"2025-09-23","cited_by_count":3,"area":"safety"},{"id":"2510.03705","title":"Backdoor-Powered Prompt Injection Attacks Nullify Defense Methods","created":"2025-10-04","cited_by_count":3,"area":"safety"},{"id":"2409.01185","title":"Backdoor Defense through Self-Supervised and Generative Learning","created":"2024-09-02","cited_by_count":2,"area":"safety"},{"id":"2409.08985","title":"Clean Label Attacks against SLU Systems","created":"2024-09-13","cited_by_count":2,"area":"safety"},{"id":"2410.21453","title":"Inverting Gradient Attacks Makes Powerful Data Poisoning","created":"2024-10-28","cited_by_count":2,"area":"safety"},{"id":"2504.05838","title":"Mind the Trojan Horse: Image Prompt Adapter Enabling Scalable and Deceptive Jailbreaking","created":"2025-04-08","cited_by_count":2,"area":"safety"},{"id":"2504.17300","title":"The Ultimate Cookbook for Invisible Poison: Crafting Subtle Clean-Label Text Backdoors with Style Attributes","created":"2025-04-24","cited_by_count":2,"area":"safety"},{"id":"2506.06003","title":"What Really is a Member? Discrediting Membership Inference via Poisoning","created":"2025-06-06","cited_by_count":2,"area":"safety"},{"id":"2511.08637","title":"How Do Data Owners Say No? A Case Study of Data Consent Mechanisms in Web-Scraped Vision-Language AI Training Datasets","created":"2025-11-10","cited_by_count":2,"area":"safety"},{"id":"2603.00516","title":"ProtegoFed: Backdoor-Free Federated Instruction Tuning with Interspersed Poisoned Data","created":"2026-02-28","cited_by_count":2,"area":"safety"},{"id":"2606.30783","title":"Security--Fidelity Tradeoffs: The Hidden Cost of Prompt Injection Defense","created":"2026-06-29","cited_by_count":2,"area":"safety"},{"id":"2607.10712","title":"Distributed Denial of Science: How Indirect Data Poisoning of AI Systems Can Industrialize Scientific Fraud","created":"2026-07-12","cited_by_count":2,"area":"safety"},{"id":"2305.09671","title":"Pick your Poison: Undetectability versus Robustness in Data Poisoning Attacks","created":"2023-05-07","cited_by_count":1,"area":"safety"},{"id":"2305.11347","title":"Quantifying the robustness of deep multispectral segmentation models against natural perturbations and data poisoning","created":"2023-05-18","cited_by_count":1,"area":"safety"},{"id":"2306.16415","title":"On Practical Aspects of Aggregation Defenses against Data Poisoning Attacks","created":"2023-06-28","cited_by_count":1,"area":"safety"},{"id":"2408.13221","title":"Protecting against simultaneous data poisoning attacks","created":"2024-08-23","cited_by_count":1,"area":"safety"},{"id":"2410.01128","title":"Using Interleaved Ensemble Unlearning to Keep Backdoors at Bay for Finetuning Vision Transformers","created":"2024-10-01","cited_by_count":1,"area":"safety"},{"id":"2412.03908","title":"Generalizable Targeted Data Poisoning against Varying Physical Objects","created":"2024-12-05","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":"2603.24857","title":"AI Security in the Foundation Model Era: A Comprehensive Survey from a Unified Perspective","created":"2026-03-25","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":"2605.19227","title":"Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models","created":"2026-05-19","cited_by_count":1,"area":"safety"},{"id":"2405.06073","title":"Hard Work Does Not Always Pay Off: Poisoning Attacks on Neural Architecture Search","created":"2024-05-09","cited_by_count":0,"area":"safety"},{"id":"2508.02312","title":"Data Security in Large Language Models: Risks, Defense, and Directions","created":"2025-08-04","cited_by_count":0,"area":"safety"},{"id":"2509.06896","title":"Are Targeted Data Poisoning Attacks as Effective as We Think?","created":"2025-09-08","cited_by_count":0,"area":"safety"},{"id":"2511.18660","title":"Subtract the Corruption: Training-Data-Free Corrective Machine Unlearning using Task Arithmetic","created":"2025-11-24","cited_by_count":0,"area":"safety"},{"id":"2512.20865","title":"Robustness Certificates for Neural Networks Against Data Poisoning and Evasion Attacks","created":"2025-12-24","cited_by_count":0,"area":"safety"},{"id":"2603.00811","title":"Curation Leaks: Membership Inference Attacks against Data Curation for Machine Learning","created":"2026-02-28","cited_by_count":0,"area":"safety"},{"id":"2603.15692","title":"BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator","created":"2026-03-16","cited_by_count":0,"area":"safety"},{"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":"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":"2605.28632","title":"Blind PRNG Hijacking: An Undetectable Integrity-Preserving Attack Against LLM Watermarking","created":"2026-05-27","cited_by_count":0,"area":"safety"},{"id":"2606.05958","title":"Steering Vectors are an Adversarial Attack Surface","created":"2026-06-04","cited_by_count":0,"area":"safety"},{"id":"2606.27511","title":"When the Aggregator Cheats: Data-Free Backdoors in Federated LLM-based QA Systems","created":"2026-06-25","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":"2607.17550","title":"(A)iSpy: Parasitic Trojans for Machine Learning Infrastructure","created":"2026-07-20","cited_by_count":0,"area":"safety"},{"id":"2608.04052","title":"When Modalities Fail to Tango: Conformal Backdoor Detection in Multimodal Contrastive Learning","created":"2026-08-04","cited_by_count":0,"area":"safety"},{"id":"2608.24354","title":"Not All Tokens Are Equal: Region-Aware Consistency Repair of Backdoors in MLLMs","created":"2026-08-25","cited_by_count":0,"area":"safety"},{"id":"2609.00351","title":"Detecting Hidden Behaviors in LLMs via Activation-matched Finetuning","created":"2026-05-29","cited_by_count":0,"area":"safety"},{"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":"2610.00663","title":"Backdoor Containment via Expert Quarantine and Shutdown in LLMs","created":"2026-09-30","cited_by_count":0,"area":"safety"},{"id":"2610.00685","title":"Backdoor Purification for LoRA-Tuned LLMs via Null-Space Projection","created":"2026-09-30","cited_by_count":0,"area":"safety"},{"id":"2610.01788","title":"SAGE: Similarity-Based Cleaning of Poisoned Training Data from Verified Examples","created":"2026-10-01","cited_by_count":0,"area":"safety"},{"id":"2610.05453","title":"The Poisoned Conversation: Privacy-Leaking Watermarks in Unified Multimodal Models","created":"2026-10-04","cited_by_count":0,"area":"safety"}],"next_cursor":null}