AI Safety

The “Knowledge–Behavior Gap” in Cultural Taboo Safety of Large Language Models ACL 2026 The Shepherd Test: How Will Super Intelligent Agents Balance Care and Control in Asymmetric Relationships? AAAI 2026 Thesis Proposal: Toward a Human-Centered and Perspective-Aware Framework for Reproducible ML Evaluation and AI Alignment ACL 2026 The Unintended Trade-off of AI Alignment: Balancing Hallucination Mitigation and Safety in LLMs EACL 2026 Too Nice to Tell the Truth: Quantifying Agreeableness-Driven Sycophancy in Role-Playing Language Models ACL 2026 Toward Controllable and Trustworthy LLM Reasoning: From Failure Mapping to Cognition-inspired Control and Real-world Impact AAAI 2026 Towards Capable and Secure Autonomous Computer-Use Agents (Student Abstract) AAAI 2026 Towards Effective, Stealthy, and Persistent Backdoor Attacks Targeting Graph Foundation Models AAAI 2026 Towards Policy-Adaptive Image Guardrail: Benchmark and Method CVPR 2026 ToxiPrompt: A Two-Stage Red-Teaming Approach for Balancing Adversarial Prompt Diversity and Response Toxicity EACL 2026 Trajectory Signatures of Deception in Large Language Models ACL 2026 Transferable Backdoor Attacks for Code Models via Sharpness-Aware Adversarial Perturbation AAAI 2026 TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy Exploration CVPR 2026 Trustworthy Autonomy Without Human Intervention in Uncertain Domains AAAI 2026 Tuning for Two Adversaries: Enhancing the Robustness Against Transfer and Query-Based Attacks Using Hyperparameter Tuning AAAI 2026 TVChain: Leveraging Textual-Visual Prompt Chains for Jailbreaking Large Vision-Language Models AAAI 2026 TWINFUZZ: Dual-Model Fuzzing for Robustness Generalization in Deep Learning AAAI 2026 UltraClean: A Simple Framework to Train Robust Neural Networks against Backdoor Attacks WACV 2026 Uncovering and Aligning Anomalous Attention Heads to Defend Against NLP Backdoor Attacks AAAI 2026 Uncovering Pretraining Code in LLMs: A Syntax-Aware Attribution Approach AAAI 2026 Understanding Emergent Misalignment via Feature Superposition Geometry ACL 2026 Understanding Jailbreak Success: A Study of Latent Space Dynamics in Large Language Models EACL 2026 Unintended Misalignment from Agentic Fine-Tuning: Risks and Mitigation AAAI 2026 Universal Adversarial Purification with DDIM Metric Loss for Stable Diffusion AAAI 2026 Unleashing the Unseen: Harnessing Benign Datasets for Jailbreaking Large Language Models EACL 2026