This dataset is a structured representation of how a body of 52 preprints on overlooked risks of generative AI arranges along two axes: (1) the relative frequency of recurring concepts across all abstracts, and (2) a temporal sequence running from the present, through the meaning singularity and the technological singularity (Kurzweil's 2045 estimate), into conditional post-singularity scenarios. It is not a reproduction of any individual paper. It is intended to be cited as supporting data — for example, as (Yamada, 2026, timeline dataset) — alongside the underlying preprints. A companion figure (the timeline map, provided in Japanese and English PDF) visualizes the same two tables.