The interactions between liquid water and hydroxyl species on Pt(111) surfaces have been intensely investigated due to their importance to fuel cell electrocatalysis. Here we present a molecular dynamics study of their structure and energetics using an ensemble of neural network potentials, which allow us to obtain unprecedented statistical sampling. We first study the energetics of hydroxyl formation, where we find a near-linear adsorption energy profile, which exhibits a soft and gradual increase in the differential adsorption energy at high hydroxyl coverages. This is strikingly different from the predictions of the conventional bilayer model, which displays a kink at 1/3ML OH coverage indicating a sizeable jump in differential adsorption energy, but within the statistical uncertainty of previously reported ab initio molecular dynamics studies. We then analyze the structure of the interface, where we provide evidence for the water–OH/Pt(111) interface being hydrophobic at high hydroxyl coverages. We furthermore explain the observed adsorption energetics by analyzing the hydrogen bonding in the water–hydroxyl adlayers, where we argue that the increase in differential adsorption energy at high OH coverage can be explained by a reduction in the number of hydrogen bonds from the adsorbed water molecules to the hydroxyls.