AI Safety

LionGuard: A Contextualized Moderation Classifier to Tackle Localized Unsafe Content COLING 2025 LiteLMGuard: Seamless and Lightweight On-Device Guardrails for Small Language Models against Quantization Vulnerabilities IJCNLP 2025 Lived Experience Not Found: LLMs Struggle to Align with Experts on Addressing Adverse Drug Reactions from Psychiatric Medication Use NAACL 2025 LLaVA-Critic: Learning to Evaluate Multimodal Models CVPR 2025 LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Models ICML 2025 LLM Agents Can Be Choice-Supportive Biased Evaluators: An Empirical Study AAAI 2025 LLM as a Broken Telephone: Iterative Generation Distorts Information ACL 2025 LlmFixer: Fix the Helpfulness of Defensive Large Language Models EMNLP 2025 LLM Jailbreak Detection for (Almost) Free! EMNLP 2025 LLM Safety for Children NAACL 2025 LLMs as Medical Safety Judges: Evaluating Alignment with Human Annotation in Patient-Facing QA ACL 2025 LLMs can be easily Confused by Instructional Distractions ACL 2025 LLMScan: Causal Scan for LLM Misbehavior Detection ICML 2025 LLMs Can Covertly Sandbag on Capability Evaluations Against Chain-of-Thought Monitoring IJCNLP 2025 LLMs Caught in the Crossfire: Malware Requests and Jailbreak Challenges ACL 2025 LLMs know their vulnerabilities: Uncover Safety Gaps through Natural Distribution Shifts ACL 2025 LLM Stinger: Jailbreaking LLMs Using RL Fine-Tuned LLMs (Student Abstract) AAAI 2025 LNE-Blocking: An Efficient Framework for Contamination Mitigation Evaluation on Large Language Models EMNLP 2025 Localizing Malicious Outputs from CodeLLM EMNLP 2025 Logicbreaks: A Framework for Understanding Subversion of Rule-based Inference ICLR 2025 LongSafety: Enhance Safety for Long-Context LLMs ACL 2025 LongSafety: Evaluating Long-Context Safety of Large Language Models ACL 2025 Long-tailed Adversarial Training with Self-Distillation ICLR 2025 Look Before You Leap: Enhance Attention and Vigilance Regarding Harmful Content with GuidelineLLM AAAI 2025 LoRATK: LoRA Once, Backdoor Everywhere in the Share-and-Play Ecosystem EMNLP 2025