This is being positioned as evidence of the Platonic Representation Hypothesis, but isn't it more likely an artifact of training models on the same/similar data

This is being positioned as evidence of the Platonic Representation Hypothesis, but isn't it more likely an artifact of training models on the same/similar data sets?