“In early 2024, the prevailing AI video generation models were limited to creating clips of approximately five seconds in length.”