ByteDance’s Seed AI lab is establishing a dedicated Model-Alignment & Preparedness department, according to an internal email reported by The Paper on October 13. Yan Lin will lead the unit and report to Seed head Wu Yonghui. Members of the existing Responsible AI team and some employees from the Chat team will transfer to the new department.
The unit will develop model-behavior specifications, safety evaluations, alignment training, protective systems and risk monitoring. Its remit includes improving AI agents’ ability to operate within human intentions and authorized limits, as well as addressing chemical, biological, radiological and nuclear risks associated with advanced models.
Seed also plans to expand public cooperation with external researchers and independent evaluators. The move creates a separate organizational home for safety and alignment work as the lab develops increasingly capable and autonomous AI systems. [The Paper, in Chinese]