AI Safety

On Prompt-Driven Safeguarding for Large Language Models ICML 2024 On scalable oversight with weak LLMs judging strong LLMs NeurIPS 2024 On the Computability of Robust PAC Learning COLT 2024 On the Convergence of an Adaptive Momentum Method for Adversarial Attacks AAAI 2024 On the Fly Neural Style Smoothing for Risk-Averse Domain Generalization WACV 2024 On the Generalization of Training-based ChatGPT Detection Methods EMNLP 2024 On the Influence of Gender and Race in Romantic Relationship Prediction from Large Language Models EMNLP 2024 On the Relationship between Truth and Political Bias in Language Models EMNLP 2024 On the Reliability of Watermarks for Large Language Models ICLR 2024 On the Scalability of Certified Adversarial Robustness with Generated Data NeurIPS 2024 On the Vulnerability of Adversarially Trained Models Against Two-faced Attacks ICLR 2024 On the Vulnerability of Safety Alignment in Open-Access LLMs ACL 2024 OODRobustBench: a Benchmark and Large-Scale Analysis of Adversarial Robustness under Distribution Shift ICML 2024 Optimal Zero-Shot Detector for Multi-Armed Attacks AISTATS 2024 Optimistic Verifiable Training by Controlling Hardware Nondeterminism NeurIPS 2024 P2BPO: Permeable Penalty Barrier-Based Policy Optimization for Safe RL AAAI 2024 PARDEN, Can You Repeat That? Defending against Jailbreaks via Repetition ICML 2024 Pareto Inverse Reinforcement Learning for Diverse Expert Policy Generation IJCAI 2024 Pedagogical Alignment of Large Language Models EMNLP 2024 Perceptions to Beliefs: Exploring Precursory Inferences for Theory of Mind in Large Language Models EMNLP 2024 Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization NeurIPS 2024 Physical Backdoor: Towards Temperature-based Backdoor Attacks in the Physical World CVPR 2024 PiShield: A PyTorch Package for Learning with Requirements IJCAI 2024 Pixel-wise Smoothing for Certified Robustness against Camera Motion Perturbations AISTATS 2024 PKAD: Pretrained Knowledge is All You Need to Detect and Mitigate Textual Backdoor Attacks EMNLP 2024