ByteDance Seed Forms AI Safety and Alignment Department

This is a structural consolidation rather than a new initiative: folding the Responsible AI team and parts of Chat into a single unit under the Seed head signals that safety and alignment are being treated as core model-development work, not a separate compliance function. The explicit mention of Agent Model safety and frontier risks like CBRN suggests ByteDance is preparing for more autonomous, higher-capability systems rather than just content moderation. The open question is whether this centralization gives the alignment team real veto power over releases or merely closer proximity to the teams it is meant to check.