AI Safety

Shifting Perspectives: Steering Vectors for Robust Bias Mitigation in LLMs EACL 2026 Should You Use LLMs to Simulate Opinions? Quality Checks for Early-Stage Deliberation AAAI 2026 Silenced Biases: The Dark Side LLMs Learned to Refuse AAAI 2026 SIRAJ: Diverse and Efficient Red-Teaming for LLM Agents via Distilled Structured Reasoning EACL 2026 SMiLE: Provably Enforcing Global Relational Properties in Neural Networks AAAI 2026 Social Dynamics as Critical Vulnerabilities that Undermine Objective Decision-Making in LLM Collectives ACL 2026 SOM Directions Are Better than One: Multi-Directional Refusal Suppression in Language Models AAAI 2026 SPA: Achieving Consensus in LLM Alignment via Self-Priority Optimization AAAI 2026 SpARK: An Embarrassingly Simple Sparse Watermarking in LLMs with Enhanced Text Quality EACL 2026 Specification-Guided Reinforcement Learning AAAI 2026 SRD: Reinforcement-Learned Semantic Perturbation for Backdoor Defense in VLMs AAAI 2026 STACK: Adversarial Attacks on LLM Safeguard Pipelines AAAI 2026 STAR-1: Safer Alignment of Reasoning LLMs with 1K Data AAAI 2026 STaR: Sensitive Trajectory Regulation for Unlearning in Large Reasoning Models AAAI 2026 Steering Safely or Off a Cliff? Rethinking Specificity and Robustness in Inference-Time Interventions EACL 2026 StyleBreak: Revealing Alignment Vulnerabilities in Large Audio-Language Models via Style-Aware Audio Jailbreak AAAI 2026 Sycophantic Anchors: Localizing and Quantifying User Agreement in Reasoning Models ACL 2026 Synthetic Text Detection in the Age of Large Language Models: Watermark vs. Automatic Detection ACL 2026 Systematic Analysis of the Unintentional CSAM-Generation-Potential of Text-to-Image Models WACV 2026 Teams of LLM Agents can Exploit Zero-Day Vulnerabilities EACL 2026 Test-time Prompt Intervention AAAI 2026 The Alignment Game: A Theory of Long-Horizon Alignment Through Recursive Curation AAAI 2026 The Doctor Will Agree With You Now: Sycophancy of Large Language Models in Multi-Turn Medical Conversations EACL 2026 The Emotional Baby Is Truly Deadly: Does Your Multimodal Large Reasoning Model Have Emotional Flattery Towards Humans? AAAI 2026 The Hidden Bias: A Study on Explicit and Implicit Political Stereotypes in Large Language Models EACL 2026