AI Safety

How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study ACL 2026 HulluEdit: Single-Pass Evidence-Consistent Subspace Editing for Mitigating Hallucinations in Large Vision-Language Models CVPR 2026 HumorReject: Decoupling LLM Safety from Refusal Prefix via a Little Humor AAAI 2026 Immunizing Models Against Harmful Long-Horizon Fine-Tuning via Contractive Optimization Dynamics CVPR 2026 Implications for AI Research: Applying Lessons from the Expert Systems Boom and Bust to the Current Large-Language Model Boom AAAI 2026 Improving CAPTCHA Robustness via Controlled Image Corruptions (Student Abstract) AAAI 2026 Inequality in the Age of Pseudonymity AAAI 2026 InEx: Hallucination Mitigation via Introspection and Cross-Modal Multi-Agent Collaboration AAAI 2026 InfrastructureSentinel: Policy Enforced Guardrails for Secure MCP-driven Infrastructure Agents AAAI 2026 Integrity Shield A System for Ethical AI Use & Authorship Transparency in Assessments EACL 2026 Interpretable Reward Model via Sparse Autoencoder AAAI 2026 Into the Gray Zone: Domain Contexts Can Blur LLM Safety Boundaries ACL 2026 Intrinsic Barriers and Practical Pathways for Human–AI Alignment: An Agreement-Based Complexity Analysis AAAI 2026 Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications ACL 2026 IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks AAAI 2026 Jailbreaking Multimodal Large Language Models using Multi-Clip Video ACL 2026 Jailbreaking Safeguarded Text-to-Image Models via Large Language Models EACL 2026 Jailbreaking Vision-Language Models via Dissonance-Guided Suffix Optimization and Image-Phrase Injection CVPR 2026 Jailbreaks as Inference-Time Alignment: A Framework for Understanding Safety Failures in LLMs EACL 2026 Jailbreak-Zero: A Path to Pareto Optimal Red Teaming for Large Language Models ACL 2026 JARVIS or Ultron? A Survey on the Safety and Security Threats of Computer-Using Agents ACL 2026 Joint-GCG: Unified Gradient-Based Poisoning Attacks on Retrieval-Augmented Generation Systems AAAI 2026 K-12EduBench: A Benchmark for Evaluating Large Language Models’ Knowledge, Problem-Solving, and Educational Goal Cognition in K-12 Education AAAI 2026 Large Language Models Struggle with Unreasonability in Math Problems AAAI 2026 LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety ACL 2026