I think you’d consider those responsible for the model to be humans, the model to be a machine, and all outputs to be derivative works of the training set with

I think you’d consider those responsible for the model to be humans, the model to be a machine, and all outputs to be derivative works of the training set with the humans holding culpability.