Key facts
- ByteDance has established a new AI safety department called Seed Model - Alignment & Preparedness.
- The unit aims to strengthen investment in AI safety and alignment.
- Its focus includes ensuring AI models align with human intentions and safety constraints.
- The department will also address risks from advanced agent models, including CBRN threats.
- Yan Lin, previously in charge of alignment for the Doubao AI model, will lead the new unit.
- Yan Lin will report to Wu Yonghui, head of ByteDance's AI model division, Seed.
ByteDance Ltd. has created a dedicated safety department within its core AI model division to manage and mitigate risks associated with artificial intelligence technology. The new unit, named Seed Model - Alignment & Preparedness, will be led by Yan Lin, who previously oversaw alignment for the company’s Doubao AI model. Yan Lin will report directly to Wu Yonghui, the head of ByteDance’s AI model division, Seed. The department's establishment is intended to strengthen investment in AI safety and alignment, ensuring that AI models behave consistently with human intentions and safety constraints. This focus aims to enhance the safety capabilities of agent models and prevent risks associated with advanced capabilities, such as those involving chemical, biological, radiological, and nuclear threats.