AI Safety

Efficient Switchable Safety Control in LLMs via Magic-Token-Guided Co-Training AAAI 2026 Efficient Verification and Falsification of ReLU Neural Barrier Certificates AAAI 2026 EigenShield: Inference-Time, Model-Agnostic Jailbreaking Defense via Causal Subspace Filtering AAAI 2026 Emergent Misalignment via In-Context Learning: Narrow in-context examples can produce broadly misaligned LLMs ACL 2026 ENCORE: Entropy-guided Reward Composition for Multi-head Safety Reward Models AAAI 2026 Enhancing All-to-X Backdoor Attacks with Optimized Target Class Mapping AAAI 2026 Enhancing Pre-training Data Detection in LLMs Through Discriminative and Symmetric Prefix Selection AAAI 2026 EquaCode: A Multi-Strategy Jailbreak Approach for Large Language Models via Equation Solving and Code Completion AAAI 2026 Error Correction in Radiology Reports: A Knowledge Distillation-Based Multi-Stage Framework AAAI 2026 Ethical Decision-making with AI: Value Alignment and the Role of Reasoning AAAI 2026 Evaluating LLMs for Police Decision-Making: A Framework Based on Police Action Scenarios AAAI 2026 Evolving Contextual Safety in Multi-Modal Large Language Models via Inference-Time Self-Reflective Memory CVPR 2026 Exploiting Synergistic Cognitive Biases to Bypass Safety in LLMs AAAI 2026 ExtendAttack: Attacking Servers of LRMs via Extending Reasoning AAAI 2026 Fact2Fiction: Targeted Poisoning Attack to Agentic Fact-checking System AAAI 2026 FACTGUARD: Event-Centric and Commonsense-Guided Fake News Detection AAAI 2026 Failures to Surface Harmful Contents in Video Large Language Models AAAI 2026 FaithLM: Towards Faithful Explanations for Large Language Models EACL 2026 FILTER: A Framework for Defending Against Backdoor Attacks in Vertical Federated Learning AAAI 2026 FINEST: Improving LLM Responses to Sensitive Topics Through Fine-Grained Evaluation EACL 2026 Fine-Tuned LLMs Know They Don’t Know: A Parameter-Efficient Approach to Recovering Honesty AAAI 2026 Formal Verification of Neural ODE for Safety Evaluation in Autonomous Vehicles AAAI 2026 FRBAT: Conditionally-Visible Physical Backdoor Attack via Fluorescence AAAI 2026 From Chaos to Cure: A Prefix Heuristics Guided Model-Agnostic Adaptive Detoxification Framework AAAI 2026 From Delegates to Trustees: How Optimizing for Long-Term Interests Shapes Bias and Alignment in LLMs EACL 2026