ByteDance Seed Creates Department Focused on AI Model Safety

ByteDance Seed has created a separate department focused on model safety and alignment, expanding its organizational focus on risks linked to increasingly capable artificial intelligence systems.

The unit, named Seed Model-Alignment & Preparedness, was announced through a recent internal email. Yan Lin will lead the department and report to Seed head Wu Yonghui.

Members of ByteDance Seed’s existing Responsible AI team and some employees from its Chat team will transfer to the new department. Its responsibilities will include aligning model behavior with human intentions and safety constraints.

The department will also work to improve safety capabilities for agent models, which can perform tasks with greater autonomy, and address risks linked to advanced chemical, biological, radiological and nuclear capabilities.