safety alignment

Safety Alignment Can Be Not Superficial With Explicit Safety Signals ICML 2025 Safety Alignment Should be Made More Than Just a Few Tokens Deep ICLR 2025 Safety Alignment via Constrained Knowledge Unlearning ACL 2025 Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety ACL 2025 SaLoRA: Safety-Alignment Preserved Low-Rank Adaptation ICLR 2025 SATA: A Paradigm for LLM Jailbreak via Simple Assistive Task Linkage ACL 2025 SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering AAAI 2025 SDD: Self-Degraded Defense against Malicious Fine-tuning ACL 2025 SDGO: Self-Discrimination-Guided Optimization for Consistent Safety in Large Language Models EMNLP 2025 SEA: Low-Resource Safety Alignment for Multimodal Large Language Models via Synthetic Embeddings ACL 2025 Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models ACL 2025 SeqAR: Jailbreak LLMs with Sequential Auto-Generated Characters NAACL 2025 Smaller Large Language Models Can Do Moral Self-Correction NAACL 2025 SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Models CVPR 2025 Speculative Safety-Aware Decoding EMNLP 2025 STAIR: Improving Safety Alignment with Introspective Reasoning ICML 2025 sudoLLM: On Multi-role Alignment of Language Models EMNLP 2025 SURE: Safety Understanding and Reasoning Enhancement for Multimodal Large Language Models EMNLP 2025 The Great Contradiction Showdown: How Jailbreak and Stealth Wrestle in Vision-Language Models? ACML 2025 The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Analysis of Orthogonal Safety Directions ICML 2025 The TIP of the Iceberg: Revealing a Hidden Class of Task-in-Prompt Adversarial Attacks on LLMs ACL 2025 Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model EMNLP 2025 Token Highlighter: Inspecting and Mitigating Jailbreak Prompts for Large Language Models AAAI 2025 Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks NAACL 2025 TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis ACL 2025