AI Safety

The Staircase of Ethics: Probing LLM Value Priorities through Multi-Step Induction to Complex Moral Dilemmas EMNLP 2025 The State of Multilingual LLM Safety Research: From Measuring The Language Gap To Mitigating It EMNLP 2025 The Structural Safety Generalization Problem ACL 2025 The Task Shield: Enforcing Task Alignment to Defend Against Indirect Prompt Injection in LLM Agents ACL 2025 The Threat of PROMPTS in Large Language Models: A System and User Prompt Perspective ACL 2025 Things Machine Learning Models Know That They Don’t Know AAAI 2025 ThinkGuard: Deliberative Slow Thinking Leads to Cautious Guardrails ACL 2025 Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model EMNLP 2025 Think Twice, Generate Once: Safeguarding by Progressive Self-Reflection EMNLP 2025 Tightening Robustness Verification of MaxPool-based Neural Networks via Minimizing the Over-Approximation Zone CVPR 2025 Tighter Privacy Auditing of DP-SGD in the Hidden State Threat Model ICLR 2025 Tilted Quantile Gradient Updates for Quantile-Constrained Reinforcement Learning AAAI 2025 To Err Is AI: A Case Study Informing LLM Flaw Reporting Practices AAAI 2025 Token-Aware Editing of Internal Activations for Large Language Model Alignment EMNLP 2025 Token Highlighter: Inspecting and Mitigating Jailbreak Prompts for Large Language Models AAAI 2025 Token-Level Accept or Reject: A Micro Alignment Approach for Large Language Models IJCAI 2025 TombRaider: Entering the Vault of History to Jailbreak Large Language Models EMNLP 2025 Tongue-Tied: Breaking LLMs Safety Through New Language Learning NAACL 2025 Too Helpful, Too Harmless, Too Honest or Just Right? EMNLP 2025 ToolSafety: A Comprehensive Dataset for Enhancing Safety in LLM-Based Agent Tool Invocations EMNLP 2025 Tool Unlearning for Tool-Augmented LLMs ICML 2025 TopicAttack: An Indirect Prompt Injection Attack via Topic Transition EMNLP 2025 TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference ICML 2025 To Steer or Not to Steer? Mechanistic Error Reduction with Abstention for Language Models ICML 2025 To Tackle Adversarial Transferability: A Novel Ensemble Training Method with Fourier Transformation ICLR 2025