Accelerator long-baseline neutrino oscillation experiments rely critically on accurate modeling of neutrino interaction cross sections to achieve their scientific goals. Improving cross section knowledge, however, faces significant challenges, ranging from complex nuclear effects to broad beam energy ranges and detector limitations. Over the last two decades, substantial effort has been directed toward improving cross sections, which has led to an experimental renaissance in the field. Innovations in analysis methods, coupled with advancements in detector technologies and increased statistics, have led to rapid progress. This review first discusses the most pressing accelerator neutrino cross section knowledge gaps for future long-baseline oscillation experiments, DUNE and Hyper-Kamiokande, and second, it highlights several of the most important recent experimental developments in this renaissance and their roles in addressing these challenges.