{"total":1055,"rows":[{"id":"2401.05566","title":"Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training","created":"2024-01-10","cited_by_count":529,"area":"agents"},{"id":"2101.05930","title":"Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks","created":"2021-01-15","cited_by_count":324,"area":null},{"id":"2302.10149","title":"Poisoning Web-Scale Training Datasets is Practical","created":"2023-02-20","cited_by_count":290,"area":null},{"id":"2402.07867","title":"PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models","created":"2024-02-12","cited_by_count":267,"area":"rag"},{"id":"2110.11571","title":"Anti-Backdoor Learning: Training Clean Models on Poisoned Data","created":"2021-10-22","cited_by_count":263,"area":null},{"id":"2110.14430","title":"Adversarial Neuron Pruning Purifies Backdoored Deep Models","created":"2021-10-27","cited_by_count":235,"area":null},{"id":"2105.12400","title":"Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger","created":"2021-05-26","cited_by_count":222,"area":"safety"},{"id":"2306.12001","title":"An Overview of Catastrophic AI Risks","created":"2023-06-21","cited_by_count":222,"area":null},{"id":"2101.04898","title":"Unlearnable Examples: Making Personal Data Unexploitable","created":"2021-01-13","cited_by_count":193,"area":null},{"id":"2110.03735","title":"Adversarial Unlearning of Backdoors via Implicit Hypergradient","created":"2021-10-07","cited_by_count":166,"area":null},{"id":"2104.03413","title":"Rethinking the Backdoor Attacks' Triggers: A Frequency Perspective","created":"2021-04-07","cited_by_count":161,"area":"safety"},{"id":"2310.19156","title":"Poisoning Retrieval Corpora by Injecting Adversarial Passages","created":"2023-10-29","cited_by_count":159,"area":null},{"id":"2202.03423","title":"Backdoor Defense via Decoupling the Training Process","created":"2022-02-05","cited_by_count":153,"area":null},{"id":"2106.09667","title":"Poisoning and Backdooring Contrastive Learning","created":"2021-06-17","cited_by_count":151,"area":null},{"id":"2503.03704","title":"Memory Injection Attacks on LLM Agents via Query-Only Interaction","created":"2025-03-05","cited_by_count":144,"area":"agents"},{"id":"2212.04612","title":"Training Data Influence Analysis and Estimation: A Survey","created":"2022-12-09","cited_by_count":135,"area":null},{"id":"2104.11315","title":"SPECTRE: Defending Against Backdoor Attacks Using Robust Statistics","created":"2021-04-22","cited_by_count":130,"area":null},{"id":"2206.12654","title":"BackdoorBench: A Comprehensive Benchmark of Backdoor Learning","created":"2022-06-25","cited_by_count":128,"area":null},{"id":"2311.14455","title":"Universal Jailbreak Backdoors from Poisoned Human Feedback","created":"2023-11-24","cited_by_count":128,"area":"alignment"},{"id":"2401.14446","title":"Black-Box Access is Insufficient for Rigorous AI Audits","created":"2024-01-25","cited_by_count":126,"area":null},{"id":"2108.00352","title":"BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning","created":"2021-08-01","cited_by_count":123,"area":"safety"},{"id":"2106.10807","title":"Adversarial Examples Make Strong Poisons","created":"2021-06-21","cited_by_count":119,"area":null},{"id":"2103.10689","title":"Interpretable Deep Learning: Interpretation, Interpretability, Trustworthiness, and Beyond","created":"2021-03-19","cited_by_count":118,"area":null},{"id":"2108.10241","title":"Back to the Drawing Board: A Critical Evaluation of Poisoning Attacks on Production Federated Learning","created":"2021-08-23","cited_by_count":118,"area":null},{"id":"2404.01099","title":"What is in Your Safe Data? Identifying Benign Data that Breaks Safety","created":"2024-04-01","cited_by_count":118,"area":"safety"},{"id":"2306.17194","title":"On the Exploitability of Instruction Tuning","created":"2023-06-28","cited_by_count":116,"area":"safety"},{"id":"2311.12075","title":"BadCLIP: Dual-Embedding Guided Backdoor Attack on Multimodal Contrastive Learning","created":"2023-11-20","cited_by_count":115,"area":"safety"},{"id":"2108.13888","title":"Backdoor Attacks on Pre-trained Models by Layerwise Weight Poisoning","created":"2021-08-31","cited_by_count":113,"area":null},{"id":"2110.07831","title":"RAP: Robustness-Aware Perturbations for Defending against Backdoor Attacks on NLP Models","created":"2021-10-15","cited_by_count":104,"area":"safety"},{"id":"2204.05255","title":"Narcissus: A Practical Clean-Label Backdoor Attack with Limited Information","created":"2022-04-11","cited_by_count":103,"area":null},{"id":"2106.08970","title":"Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratch","created":"2021-06-16","cited_by_count":102,"area":"safety"},{"id":"2206.10341","title":"Neurotoxin: Durable Backdoors in Federated Learning","created":"2022-06-12","cited_by_count":98,"area":null},{"id":"2307.14692","title":"Backdoor Attacks for In-Context Learning with Language Models","created":"2023-07-27","cited_by_count":96,"area":"safety"},{"id":"2403.13355","title":"BadEdit: Backdooring large language models by model editing","created":"2024-03-20","cited_by_count":96,"area":"safety"},{"id":"2205.01992","title":"Wild Patterns Reloaded: A Survey of Machine Learning Security against Training Data Poisoning","created":"2022-05-04","cited_by_count":94,"area":"safety"},{"id":"2107.06641","title":"Trustworthy AI: A Computational Perspective","created":"2021-07-12","cited_by_count":93,"area":null},{"id":"2106.06361","title":"Turn the Combination Lock: Learnable Textual Backdoor Attacks via Word Substitution","created":"2021-06-11","cited_by_count":89,"area":"safety"},{"id":"2303.05762","title":"TrojDiff: Trojan Attacks on Diffusion Models with Diverse Targets","created":"2023-03-10","cited_by_count":88,"area":"diffusion"},{"id":"2403.05030","title":"Defending Against Unforeseen Failure Modes with Latent Adversarial Training","created":"2024-03-08","cited_by_count":88,"area":"safety"},{"id":"2402.13851","title":"VL-Trojan: Multimodal Instruction Backdoor Attacks against Autoregressive Visual Language Models","created":"2024-02-21","cited_by_count":87,"area":"mllm"},{"id":"2205.13383","title":"BppAttack: Stealthy and Efficient Trojan Attacks against Deep Neural Networks via Image Quantization and Contrastive Adversarial Learning","created":"2022-05-26","cited_by_count":86,"area":null},{"id":"2210.00875","title":"Untargeted Backdoor Watermark: Towards Harmless and Stealthy Dataset Copyright Protection","created":"2022-09-27","cited_by_count":86,"area":null},{"id":"2110.02467","title":"BadPre: Task-agnostic Backdoor Attacks to Pre-trained NLP Foundation Models","created":"2021-10-06","cited_by_count":85,"area":"safety"},{"id":"2104.02361","title":"Backdoor Attack in the Physical World","created":"2021-04-06","cited_by_count":84,"area":null},{"id":"2201.06640","title":"Towards Adversarial Evaluations for Inexact Machine Unlearning","created":"2022-01-17","cited_by_count":84,"area":"safety"},{"id":"2201.08135","title":"Survey on Federated Learning Threats: concepts, taxonomy on attacks and defences, experimental study and challenges","created":"2022-01-20","cited_by_count":84,"area":null},{"id":"2206.04920","title":"Fisher SAM: Information Geometry and Sharpness Aware Minimisation","created":"2022-06-10","cited_by_count":84,"area":null},{"id":"2401.12242","title":"BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models","created":"2024-01-20","cited_by_count":80,"area":"lr"},{"id":"2406.20053","title":"Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptation","created":"2024-06-28","cited_by_count":80,"area":null},{"id":"2208.03111","title":"Data-free Backdoor Removal based on Channel Lipschitzness","created":"2022-08-05","cited_by_count":79,"area":null},{"id":"2209.08773","title":"CATER: Intellectual Property Protection on Text Generation APIs via Conditional Watermarks","created":"2022-09-19","cited_by_count":76,"area":"safety"},{"id":"2104.09667","title":"Manipulating SGD with Data Ordering Attacks","created":"2021-04-19","cited_by_count":74,"area":"safety"},{"id":"2206.08514","title":"A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarks","created":"2022-06-17","cited_by_count":74,"area":"safety"},{"id":"2103.13127","title":"Black-box Detection of Backdoor Attacks with Limited Information and Data","created":"2021-03-24","cited_by_count":73,"area":null},{"id":"2310.14735","title":"Unleashing the potential of prompt engineering for large language models","created":"2023-10-23","cited_by_count":72,"area":null},{"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":"2305.00054","title":"LAVA: Data Valuation without Pre-Specified Learning Algorithms","created":"2023-04-28","cited_by_count":69,"area":null},{"id":"2105.00579","title":"BACKDOORL: Backdoor Attack against Competitive Reinforcement Learning","created":"2021-05-02","cited_by_count":68,"area":null},{"id":"2102.09159","title":"Robust and Differentially Private Mean Estimation","created":"2021-02-18","cited_by_count":66,"area":null},{"id":"2305.10862","title":"How Deep Learning Sees the World: A Survey on Adversarial Attacks & Defenses","created":"2023-05-18","cited_by_count":66,"area":null},{"id":"2209.06015","title":"Black-box Dataset Ownership Verification via Backdoor Watermarking","created":"2022-08-04","cited_by_count":65,"area":null},{"id":"2304.11823","title":"Enhancing Fine-Tuning Based Backdoor Defense with Sharpness-Aware Minimization","created":"2023-04-24","cited_by_count":64,"area":null},{"id":"2106.04690","title":"Handcrafted Backdoors in Deep Neural Networks","created":"2021-06-08","cited_by_count":63,"area":null},{"id":"2209.11715","title":"The \"Beatrix'' Resurrections: Robust Backdoor Detection via Gram Matrices","created":"2022-09-23","cited_by_count":63,"area":null},{"id":"2303.03323","title":"CleanCLIP: Mitigating Data Poisoning Attacks in Multimodal Contrastive Learning","created":"2023-03-06","cited_by_count":63,"area":"safety"},{"id":"2112.01148","title":"FIBA: Frequency-Injection based Backdoor Attack in Medical Image Analysis","created":"2021-12-02","cited_by_count":61,"area":null},{"id":"2101.11073","title":"Property Inference From Poisoning","created":"2021-01-26","cited_by_count":60,"area":null},{"id":"2301.02344","title":"TrojanPuzzle: Covertly Poisoning Code-Suggestion Models","created":"2023-01-06","cited_by_count":60,"area":null},{"id":"2102.04716","title":"Better Safe Than Sorry: Preventing Delusive Adversaries with Adversarial Training","created":"2021-02-09","cited_by_count":59,"area":null},{"id":"2204.06974","title":"Planting Undetectable Backdoors in Machine Learning Models","created":"2022-04-14","cited_by_count":59,"area":null},{"id":"2211.02408","title":"Rickrolling the Artist: Injecting Backdoors into Text Encoders for Text-to-Image Synthesis","created":"2022-11-04","cited_by_count":59,"area":"safety"},{"id":"2211.08095","title":"Will Large-scale Generative Models Corrupt Future Datasets?","created":"2022-11-15","cited_by_count":59,"area":null},{"id":"2305.14876","title":"Reconstructive Neuron Pruning for Backdoor Defense","created":"2023-05-24","cited_by_count":59,"area":null},{"id":"2210.09545","title":"Fine-mixing: Mitigating Backdoors in Fine-tuned Language Models","created":"2022-10-18","cited_by_count":57,"area":"safety"},{"id":"2310.14942","title":"Domain Watermark: Effective and Harmless Dataset Copyright Protection is Closed at Hand","created":"2023-10-09","cited_by_count":57,"area":null},{"id":"2205.06900","title":"MM-BD: Post-Training Detection of Backdoor Attacks with Arbitrary Backdoor Pattern Types Using a Maximum Margin Statistic","created":"2022-05-13","cited_by_count":55,"area":null},{"id":"2303.15180","title":"Detecting Backdoors in Pre-trained Encoders","created":"2023-03-23","cited_by_count":55,"area":"safety"},{"id":"2311.16153","title":"Identifying and Mitigating Vulnerabilities in LLM-Integrated Applications","created":"2023-11-07","cited_by_count":54,"area":null},{"id":"2110.14880","title":"AEVA: Black-box Backdoor Detection Using Adversarial Extreme Value Analysis","created":"2021-10-28","cited_by_count":53,"area":null},{"id":"2204.05011","title":"FederatedScope: A Flexible Federated Learning Platform for Heterogeneity","created":"2022-04-11","cited_by_count":53,"area":null},{"id":"2310.16608","title":"Performative Prediction: Past and Future","created":"2023-10-25","cited_by_count":53,"area":null},{"id":"2109.08266","title":"Hard to Forget: Poisoning Attacks on Certified Machine Unlearning","created":"2021-09-17","cited_by_count":52,"area":"safety"},{"id":"2202.02628","title":"Improved Certified Defenses against Data Poisoning with (Deterministic) Finite Aggregation","created":"2022-02-05","cited_by_count":52,"area":"safety"},{"id":"2108.02488","title":"Poison Ink: Robust and Invisible Backdoor Attack","created":"2021-08-05","cited_by_count":50,"area":null},{"id":"2206.10013","title":"Measuring the Effect of Training Data on Deep Learning Predictions via Randomized Experiments","created":"2022-06-20","cited_by_count":50,"area":null},{"id":"2101.06969","title":"Red Alarm for Pre-trained Models: Universal Vulnerability to Neuron-Level Backdoor Attacks","created":"2021-01-18","cited_by_count":49,"area":null},{"id":"2301.02496","title":"Stealthy Backdoor Attack for Code Models","created":"2023-01-06","cited_by_count":49,"area":"safety"},{"id":"2406.18844","title":"Revisiting Backdoor Attacks against Large Vision-Language Models from Domain Shift","created":"2024-06-27","cited_by_count":49,"area":"mllm"},{"id":"2502.08586","title":"Commercial LLM Agents Are Already Vulnerable to Simple Yet Dangerous Attacks","created":"2025-02-12","cited_by_count":49,"area":"agents"},{"id":"2205.08305","title":"A Study of the Attention Abnormality in Trojaned BERTs","created":"2022-05-13","cited_by_count":48,"area":null},{"id":"2306.06123","title":"Adversarial attacks and defenses in explainable artificial intelligence: A survey","created":"2023-06-06","cited_by_count":48,"area":null},{"id":"2410.13722","title":"Persistent Pre-Training Poisoning of LLMs","created":"2024-10-17","cited_by_count":48,"area":null},{"id":"2110.12925","title":"CoProtector: Protect Open-Source Code against Unauthorized Training Usage with Data Poisoning","created":"2021-10-25","cited_by_count":47,"area":"safety"},{"id":"2202.05749","title":"Constrained Optimization with Dynamic Bound-scaling for Effective NLPBackdoor Defense","created":"2022-02-11","cited_by_count":47,"area":"safety"},{"id":"2112.06274","title":"SparseFed: Mitigating Model Poisoning Attacks in Federated Learning with Sparsification","created":"2021-12-12","cited_by_count":46,"area":null},{"id":"2205.12709","title":"VeriFi: Towards Verifiable Federated Unlearning","created":"2022-05-25","cited_by_count":46,"area":"safety"},{"id":"2206.03693","title":"Autoregressive Perturbations for Data Poisoning","created":"2022-06-08","cited_by_count":46,"area":"safety"},{"id":"2303.18191","title":"Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistency","created":"2023-03-27","cited_by_count":46,"area":null},{"id":"2304.02786","title":"UNICORN: A Unified Backdoor Trigger Inversion Framework","created":"2023-04-05","cited_by_count":46,"area":null},{"id":"2106.14851","title":"Data Poisoning Won't Save You From Facial Recognition","created":"2021-06-28","cited_by_count":45,"area":"safety"}],"next_cursor":"100"}