AI Safety

Shh, don't say that! Domain Certification in LLMs ICLR 2025 ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning ICML 2025 ShieldHead: Decoding-time Safeguard for Large Language Models ACL 2025 Shield Synthesis for LTL Modulo Theories AAAI 2025 Shortcut Learning in Safety: The Impact of Keyword Bias in Safeguards ACL 2025 SilverSpeak: Evading AI-Generated Text Detectors using Homoglyphs COLING 2025 SIMBA UQ: Similarity-Based Aggregation for Uncertainty Quantification in Large Language Models EMNLP 2025 Simulate and Eliminate: Revoke Backdoors for Generative Large Language Models AAAI 2025 SimVBG: Simulating Individual Values by Backstory Generation EMNLP 2025 Single Character Perturbations Break LLM Alignment AAAI 2025 Small Changes, Big Impact: How Manipulating a Few Neurons Can Drastically Alter LLM Aggression ACL 2025 Smaller Large Language Models Can Do Moral Self-Correction NAACL 2025 SMLE: Safe Machine Learning via Embedded Overapproximation AAAI 2025 Soft Token Attacks Cannot Reliably Audit Unlearning in Large Language Models EMNLP 2025 SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal ICLR 2025 SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Models CVPR 2025 SPD: Shallow Backdoor Protecting Deep Backdoor Against Backdoor Detection ICCV 2025 Speculative Safety-Aware Decoding EMNLP 2025 Speed Master: Quick or Slow Play to Attack Speaker Recognition AAAI 2025 SPICA: Retrieving Scenarios for Pluralistic In-Context Alignment ACL 2025 SPIRIT: Patching Speech Language Models against Jailbreak Attacks EMNLP 2025 SPoRt - Safe Policy Ratio: Certified Training and Deployment of Task Policies in Model-Free RL IJCAI 2025 SQL Injection Jailbreak: A Structural Disaster of Large Language Models ACL 2025 SSNTrio@DravidianLangTech 2025: Identification of AI Generated Content in Dravidian Languages using Transformers NAACL 2025 STAIR: Improving Safety Alignment with Introspective Reasoning ICML 2025