AI Safety

Imitate the Good and Avoid the Bad: An Incremental Approach to Safe Reinforcement Learning AAAI 2024 Immunization against harmful fine-tuning attacks EMNLP 2024 Impeding LLM-assisted Cheating in Introductory Programming Assignments via Adversarial Perturbation EMNLP 2024 Imperio: Language-Guided Backdoor Attacks for Arbitrary Model Control IJCAI 2024 Implicit Personalization in Language Models: A Systematic Study EMNLP 2024 Improved Generation of Adversarial Examples Against Safety-aligned LLMs NeurIPS 2024 Improving Alignment and Robustness with Circuit Breakers NeurIPS 2024 Improving Generalization of Alignment with Human Preferences through Group Invariant Learning ICLR 2024 Inconsistent dialogue responses and how to recover from them EACL 2024 INDICT: Code Generation with Internal Dialogues of Critiques for Both Security and Helpfulness NeurIPS 2024 InferAligner: Inference-Time Alignment for Harmlessness through Cross-Model Guidance EMNLP 2024 Information-Theoretic Opacity-Enforcement in Markov Decision Processes IJCAI 2024 InfoRM: Mitigating Reward Hacking in RLHF via Information-Theoretic Reward Modeling NeurIPS 2024 InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents ACL 2024 Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models NeurIPS 2024 Inspecting Prediction Confidence for Detecting Black-Box Backdoor Attacks AAAI 2024 Instructions as Backdoors: Backdoor Vulnerabilities of Instruction Tuning for Large Language Models NAACL 2024 Instruction Tuning for Secure Code Generation ICML 2024 Intent-Aware and Hate-Mitigating Counterspeech Generation via Dual-Discriminator Guided LLMs COLING 2024 Interactive Theorem Provers: Applications in AI, Opportunities, and Challenges AAAI 2024 Interpretable Concept Bottlenecks to Align Reinforcement Learning Agents NeurIPS 2024 Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts EMNLP 2024 Interpreting Robustness Proofs of Deep Neural Networks ICLR 2024 Intersecting-Boundary-Sensitive Fingerprinting for Tampering Detection of DNN Models ICML 2024 Intersectionality in AI Safety: Using Multilevel Models to Understand Diverse Perceptions of Safety in Conversational AI COLING 2024