ByteDance has created a dedicated safety and alignment department within its AI research division Seed. This new structure, called Model-Alignment & Preparedness, will start with ten researchers reporting directly to Seed’s leader Wu Yonghui, under the supervision of Yan Lin. Positions are based in Beijing, Shanghai, and San Jose, California. The department will focus on model behavior specifications, safety evaluations, automated red-teaming, and CBRN preparedness covering chemical, biological, radiological, and nuclear threats. This reorganization marks the first time a Chinese AI lab has created a standalone safety and alignment department, elevating these functions to the same level as model capability research.
Source: Read the original article
