ByteDance Seed establishes the Model-Alignment and Preparedness department led by Yan Lin to enhance AI safety. This new unit focuses on securing intelligent agent models and mitigating CBRN risks through regulation, evaluation, and alignment training. As the first Chinese frontier AI lab to create a dedicated safety department, Seed aims to ensure models operate within safe constraints, reducing misuse and behavioral deviation while promoting open collaboration with external researchers to support large-scale application.