AI Safety

Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models? CVPR 2025 DREAM: Disentangling Risks to Enhance Safety Alignment in Multimodal Large Language Models NAACL 2025 DROWN: Towards Tighter LiRPA-based Robustness Certification COLING 2025 DSCD: Large Language Model Detoxification with Self-Constrained Decoding EMNLP 2025 Durable Quantization Conditioned Misalignment Attack on Large Language Models ICLR 2025 DyePack: Provably Flagging Test Set Contamination in LLMs Using Backdoors EMNLP 2025 Dynamic Evaluation for Oversensitivity in LLMs EMNLP 2025 Dynamic Evaluation with Cognitive Reasoning for Multi-turn Safety of Large Language Models ACL 2025 Dynamic Evil Score-Guided Decoding: An Efficient Decoding Framework For Red-Team Model ACL 2025 Dynamic Guided and Domain Applicable Safeguards for Enhanced Security in Large Language Models NAACL 2025 EAVIT: Efficient and Accurate Human Value Identification From Text Data via LLMs IJCAI 2025 Echoes of Agreement: Argument Driven Sycophancy in Large Language models EMNLP 2025 Effective Red-Teaming of Policy-Adherent Agents EMNLP 2025 Efficient but Vulnerable: Benchmarking and Defending LLM Batch Prompting Attack ACL 2025 EFFICIENT JAILBREAK ATTACK SEQUENCES ON LARGE LANGUAGE MODELS VIA MULTI-ARMED BANDIT-BASED CONTEXT SWITCHING ICLR 2025 Efficient Robustness Evaluation via Constraint Relaxation AAAI 2025 Efficient Safety Alignment of Large Language Models via Preference Re-ranking and Representation-based Reward Modeling ACL 2025 EIA: ENVIRONMENTAL INJECTION ATTACK ON GENERALIST WEB AGENTS FOR PRIVACY LEAKAGE ICLR 2025 ELAB: Extensive LLM Alignment Benchmark in Persian Language ACL 2025 ELBA-Bench: An Efficient Learning Backdoor Attacks Benchmark for Large Language Models ACL 2025 Eliciting Human Preferences with Language Models ICLR 2025 ELITE: Enhanced Language-Image Toxicity Evaluation for Safety ICML 2025 Embedding Safety into RL: A New Take on Trust Region Methods ICML 2025 Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs ICML 2025 Emerging Safety Attack and Defense in Federated Instruction Tuning of Large Language Models ICLR 2025