AI Safety

Learning and Forgetting Unsafe Examples in Large Language Models ICML 2024 Learning Human-like Representations to Enable Learning Human Values NeurIPS 2024 Learning Performance-oriented Control Barrier Functions Under Complex Safety Constraints and Limited Actuation CoRL 2024 Learning Safety Constraints from Demonstrations with Unknown Rewards AISTATS 2024 Learning To See But Forgetting To Follow: Visual Instruction Tuning Makes LLMs More Prone To Jailbreak Attacks COLING 2024 Learn to Refuse: Making Large Language Models More Controllable and Reliable through Knowledge Scope Limitation and Refusal Mechanism EMNLP 2024 Let Me Teach You: Pedagogical Foundations of Feedback for Language Models EMNLP 2024 Linguistic Obfuscation Attacks and Large Language Model Uncertainty EACL 2024 Linguistic Rule Induction Improves Adversarial and OOD Robustness in Large Language Models COLING 2024 LIRE: listwise reward enhancement for preference alignment ACL 2024 Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack NeurIPS 2024 LLM Evaluators Recognize and Favor Their Own Generations NeurIPS 2024 LLMGuard: Guarding against Unsafe LLM Behavior AAAI 2024 LLM Internal States Reveal Hallucination Risk Faced With a Query EMNLP 2024 Long-Term Safe Reinforcement Learning with Binary Feedback AAAI 2024 Look Who’s Talking Now: Covert Channels From Biased LLMs EMNLP 2024 LOTUS: Evasive and Resilient Backdoor Attacks through Sub-Partitioning CVPR 2024 LT-Defense: Searching-free Backdoor Defense via Exploiting the Long-tailed Effect NeurIPS 2024 Lyapunov-Stable Deep Equilibrium Models AAAI 2024 Making Harmful Behaviors Unlearnable for Large Language Models ACL 2024 Many-shot Jailbreaking NeurIPS 2024 MAP: MAsk-Pruning for Source-Free Model Intellectual Property Protection CVPR 2024 MART: Improving LLM Safety with Multi-round Automatic Red-Teaming NAACL 2024 MathAttack: Attacking Large Language Models towards Math Solving Ability AAAI 2024 Measuring Goal-Directedness NeurIPS 2024