Does this AI veganism extend to every model with training data that is not in the public domain? Every language model, every translation model, every classific

Does this AI veganism extend to every model with training data that is not in the public domain? Every language model, every translation model, every classification and detection model build on top of Imagenet weights?