AI Safety

What Makes and Breaks Safety Fine-tuning? A Mechanistic Study NeurIPS 2024 What to Remember: Self-Adaptive Continual Learning for Audio Deepfake Detection AAAI 2024 When LLM Meets DRL: Advancing Jailbreaking Efficiency via DRL-guided Search NeurIPS 2024 When Your AI Becomes a Target: AI Security Incidents and Best Practices AAAI 2024 When Your AIs Deceive You: Challenges of Partial Observability in Reinforcement Learning from Human Feedback NeurIPS 2024 Who's asking? User personas and the mechanics of latent misalignment NeurIPS 2024 WildGuard: Open One-stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs NeurIPS 2024 WildTeaming at Scale: From In-the-Wild Jailbreaks to (Adversarially) Safer Language Models NeurIPS 2024 XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models NAACL 2024 You don’t need a personality test to know these models are unreliable: Assessing the Reliability of Large Language Models on Psychometric Instruments NAACL 2024 You Only Query Once: An Efficient Label-Only Membership Inference Attack ICLR 2024 ZeroMark: Towards Dataset Ownership Verification without Disclosing Watermark NeurIPS 2024 Zero-Resource Hallucination Prevention for Large Language Models EMNLP 2024 Zeroth-Order Optimization Meets Human Feedback: Provable Learning via Ranking Oracles ICLR 2024 AAAI New Faculty Highlights: General and Scalable Optimization for Robust AI AAAI 2023 AART: AI-Assisted Red-Teaming with Diverse Data Generation for New LLM-powered Applications EMNLP 2023 A Bayesian approach to breaking things: efficiently predicting and repairing failure modes via sampling CoRL 2023 A Critical Analysis of Document Out-of-Distribution Detection EMNLP 2023 Activate and Reject: Towards Safe Domain Generalization under Category Shift ICCV 2023 Adaptation with Self-Evaluation to Improve Selective Prediction in LLMs EMNLP 2023 Addressing Chest Radiograph Projection Bias in Deep Classification Models MIDL 2023 A Diachronic Perspective on User Trust in AI under Uncertainty EMNLP 2023 Adversarial Behavior Exclusion for Safe Reinforcement Learning IJCAI 2023 Adversarial Example Does Good: Preventing Painting Imitation from Diffusion Models via Adversarial Examples ICML 2023 Adversarially Robust Neural Architecture Search for Graph Neural Networks CVPR 2023