The transformation of ozone (O3) into hydroxyl radical (•OH) during the ozonation was evaluated in the presence of granular activated carbon (GAC) and biofilm-covered granular activated carbon (BGAC). While both GAC and BGAC accelerated O3 decomposition, the manner in which they generated •OH was different. GAC catalyzed the conversion of O3 into •OH, with increasing hydroxyl radical exposure (∫[•OH]dt) by 24 −72 % depending on the GAC dose. Conversely, BGAC exhibited limited capacity for •OH generation from O3 decomposition, resulting in decreased ∫[•OH]dt by 14 −25 % at higher BGAC doses. This disparity is likely attributable to biofilms on BGAC surface obstructing catalytic sites and hinder O3-to-•OH conversion. These different behaviors influenced the degradation of contaminants during ozonation. Specifically, GAC enhanced the degradation of O3-resistant contaminants, whereas BGAC inhibited it. Machine learning (ML) models were developed based on experimental data to predict oxidant exposures (R2 > 0.99 for ∫[O3]dt and ∫[•OH]dt). The elimination of several contaminants in systems was successfully predicted by these ML models, coupled with a straightforward kinetic equation that included adsorption and oxidation parameters. Additionally, ozonation modified the catalytic properties of GAC and BGAC. Extended ozonation oxidized GAC surface, diminishing its capability to convert O3 into •OH. In contrast, oxidation of BGAC disrupted surface biofilms, thereby restoring its catalytic function.