AI safety papers at major conferences
Total 69 papers
- Adversarial Observations in Weather ForecastingErik Imgrund, Thorsten Eisenhofer, Konrad RieckConference page
- Busting the Paper Ballot: Voting Meets Adversarial Machine LearningKaleel Mahmood, Caleb Manicke, Ethan Rathbun et al.Conference page
- Evaluating the robustness of a production malware detection system to transferable adversarial attacksMilad Nasr, Yanick Fratantonio, Luca Invernizzi et al.Conference page
- GASLITEing the Retrieval: Exploring Vulnerabilities in Dense Embedding-based SearchMatan Ben-Tov, Mahmood SharifConference page
- One Surrogate to Fool Them All: Universal, Transferable, and Targeted Adversarial Attacks with CLIPBinyan Xu, Xilin Dai, Di Tang et al.Conference page
Prompt injection
3- FlippedRAG: Black-Box Opinion Manipulation Attacks to Retrieval-Augmented Generation of Large Language ModelsZhuo Chen, Yuyang Gong, Miaokun Chen et al.Conference page
- Here Comes The AI Worm: Preventing the Propagation of Adversarial Self-Replicating Prompts Within GenAI EcosystemsStav Cohn, Ron Bitton, Nassi BenConference page
- SecAlign: Defending Against Prompt Injection with Preference OptimizationSizhe Chen, Arman Zharmagambetov, Saeed Mahloujifar et al.Conference page
- AgentSentinel: An End-to-End and Real-Time Security Defense Framework for Computer-Use AgentsHaitao Hu, Peng Chen, Yanpeng Zhao et al.Conference page
- Adversarially Robust Assembly Language Model for Packed Executables Detection
- Analyzing PDFs like Binaries: Adversarially Robust PDF Malware Analysis via Intermediate Representation and Language Model
- Deep Learning from Imperfectly Labeled Malware DataFahad Alotaibi, Euan Goodbrand, Sergio MaffeisConference page
- Exact Robustness Certification of k-Nearest NeighborsFrancesco Ranzato, Ahmad Shakeel, Marco ZanellaConference page
- Give LLMs a Security Course: Securing Retrieval-Augmented Code Generation via Knowledge Injection
- Hardening Deep Neural Network Binaries against Reverse Engineering Attacks
- Provable Repair of Deep Neural Network Defects by Preimage Synthesis and Property RefinementJianan Ma, Jingyi Wang, Qi Xuan et al.Conference page
- RAG-WM: An Efficient Black-Box Watermarking Approach for Retrieval-Augmented Generation of Large Language ModelsPeizhuo Lv, Mengjie Sun, Hao Wang et al.Conference page
- Removal Attack and Defense on AI Generated Content Latent-based WatermarkingDe Zhang Lee, Han Fang, Hanyi Wang et al.Conference page
- Rethinking Machine Unlearning in Image Generation ModelsRenyang Liu, Wenjie Feng, Tianwei Zhang et al.Conference page
- SafeGuider: Robust and Practical Content Safety Control for Text-to-Image ModelsPeigui Qi, Kunsheng Tang, Wenbo Zhou et al.Conference page
- Sylva: Tailoring Personalized Adversarial Defense in Pre-trained Models via Collaborative Fine-tuningTianyu Qi, Lei Xue, Yufeng Zhan et al.Conference page
- UnsafeBench: Benchmarking Image Safety Classifiers on Real-World and AI-Generated Images
- YouthSafe: A Youth-Centric Safety Benchmark and Safeguard Model for Large Language Models
- A Practical and Secure Byzantine Robust AggregatorDe Zhang Lee, Aashish Kolluri, Prateek Saxena et al.Conference page
- Cascading Adversarial Bias from Injection to Distillation in Language ModelsHarsh Chaudhari, Jamie Hayes, Matthew Jagielski et al.Conference page
- FilterFL: Knowledge Filtering-based Data-Free Backdoor Defense for Federated LearningYanxin Yang, Ming Hu, Xiaofei Xie et al.Conference page
- On Hyperparameters and Backdoor-Resistance in Horizontal Federated Learning
- On the Feasibility of Poisoning Text-to-Image AI Models via Adversarial MislabelingStanley Wu, Ronik Bhaskar, Anna Yoo Jeong Ha et al.Conference page
- PoisonSpot: Precise Spotting of Clean-Label Backdoors via Fine-Grained Training Provenance TrackingPhilemon Hailemariam, Birhanu EsheteConference page
- The Phantom Menace in PET-Hardened Deep Learning Models: Invisible Configuration-Induced AttacksYiteng Peng, Dongwei Xiao, Zhibo Liu et al.Conference page
- Towards Backdoor Stealthiness in Model Parameter SpaceXiaoyun Xu, Zhuoran Liu, Stefanos Koffas et al.Conference page
- VillainNet: Targeted Poisoning Attacks Against SuperNets Along the Accuracy-Latency Pareto FrontierDavid Oygenblik, Abhinav Vemulapalli, Animesh Agrawal et al.Conference page
- What’s Pulling the Strings? Evaluating Integrity and Attribution in AI Training and Inference through Concept ShiftJiamin Chang, Haoyang Li, Hammond Pearce et al.Conference page
Privacy and data breaches
27- Anonymity Unveiled: A Practical Framework for Auditing Data Use in Deep Learning Models
- Armadillo: Robust Single-Server Secure Aggregation for Federated Learning with Input Validation
- Can Personal Health Information Be Secured in LLM? Privacy Attack and Defense in the Medical Domain
- Co-Prime: A Co-design Framework for Privacy Preserving Machine Learning on FPGAShuo Xu, Jiming Xu, Pengfei Xue et al.Conference page
- CryptGNN: Enabling Secure Inference for Graph Neural NetworksPritam Sen, Yao Ma, Cristian BorceaConference page
- DCMI: A Differential Calibration Membership Inference Attack Against Retrieval-Augmented Generation
- Differentiation-Based Extraction of Proprietary Data from Fine-tuned LLMsZongjie Li, Daoyuan Wu, Shuai Wang et al.Conference page
- DivTrackee versus DynTracker: Promoting Diversity in Anti-Facial Recognition against Dynamic FR StrategyWenshu Fan, Minxing Zhang, Hongwei Li et al.Conference page
- DPImageBench: A Unified Benchmark for Differentially Private Image SynthesisChen GONG, Kecen Li, Zinan Lin et al.Conference page
- Harnessing Sparsification in Federated Learning: A Secure, Efficient, and Differentially Private Realization
- Membership Inference Attacks as Privacy Tools: Reliability, Disparity and EnsembleZhiqi Wang, Chengyu Zhang, Yuetian Chen et al.Conference page
- MoEcho: Exploiting Side-Channel Attacks to Compromise User Privacy in Mixture-of-Experts LLMsRuyi Ding, Tianhong Xu, Xinyi Shen et al.Conference page
- Mosformer: Maliciously Secure Three-Party Inference Framework for Large TransformersKe Cheng, Yuheng Xia, Anxiao Song et al.Conference page
- PLRV-O: Advancing Differentially Private Deep Learning via Privacy Loss Random Variable OptimizationQin Yang, Nicholas Stout, Meisam Mohammady et al.Conference page
- PreferCare: Preference Dataset Copyright Protection in LLM Alignment by Watermark Injection and VerificationJian Lou, Chenyang Zhang, Xiaoyu Zhang et al.Conference page
- Prompt Inference Attack on Distributed Large Language Model Inference FrameworksXinjian Luo, Ting Yu, Xiaokui XiaoConference page
- Prototype Surgery: Tailoring Neural Prototypes via Soft Labels for Efficient Machine UnlearningGaoyang Liu, Xijie Wang, Zixiong Wang et al.Conference page
- Riddle Me This! Stealthy Membership Inference for Retrieval-Augmented Generation
- Safeguarding Graph Neural Networks against Topology Inference AttacksJie Fu, Yuan Hong, Zhili Chen et al.Conference page
- SCOPE: Expanding Client-Side Post-Processing for Efficient Privacy-Preserving Model InferenceShenchen Zhu, Kai Chen, Yue Zhao et al.Conference page
- Split UnlearningYanna Jiang, Guangsheng Yu, QIN WANG et al.Conference page
- TensorShield: Safeguarding On-Device Inference by Shielding Critical DNN Tensors with TEETong Sun, Bowen Jiang, Hailong Lin et al.Conference page
- THOR: Secure Transformer Inference with Homomorphic EncryptionJungho Moon, Dongwoo Yoo, Xiaoqian Jiang et al.Conference page
- WPC: Weight Plaintext Compression for CNN Inference based on RNS-CKKSGuiming Shi, Yuchen Wei, Shengyu Fan et al.Conference page
- You Can't Steal Nothing: Mitigating Prompt Leakages in LLMs via System VectorsBochuan Cao, Yuanpu Cao, Changjiang Li et al.Conference page
- Zero-Knowledge AI Inference with High Precision
- ZORRO: Zero-Knowledge Robustness and Privacy for Split LearningNojan Sheybani, Alessandro Pegoraro, Jonathan Knauer et al.Conference page
Supply chain security
4- ImportSnare: Directed ”Code Manual” Hijacking in Retrieval-Augmented Code GenerationKai Ye, Liangcai Su, Chenxiong QianConference page
- PickleBall: Secure Deserialization of Pickle-based Machine Learning Models
- Sentry: Authenticating Machine Learning Artifacts on the FlyAndrew Gan, Zahra GhodsiConference page
- What Lurks Within? Concept Auditing for Shared Diffusion Models at ScaleXiaoyong Yuan, Xiaolong Ma, Linke Guo et al.Conference page
AI and cyber offense and defense
5- JsDeObsBench: Measuring and Benchmarking LLMs for JavaScript DeobfuscationGuoqiang Chen, Xin Jin, Zhiqiang LinConference page
- Measuring and Augmenting Large Language Models for Solving Capture-the-Flag Challenges
- OCR-APT: Reconstructing APT Stories from Audit Logs using Subgraph Anomaly Detection and LLMs
- Oedipus: LLM-enchanced Reasoning CAPTCHA SolverGelei Deng, Haoran Ou, Yi Liu et al.Conference page
- PromeFuzz: A Knowledge-Driven Approach to Fuzzing Harness Generation with Large Language Models