Our paper titled EASE: Practical and Efficient Safety Alignment for Small Language Models was accepted by AAAI 2026.