How will the EU enforce this? Will they go through the training dataset of each company’s AI models? Also, given that training datasets are closed source, there’s practically no way to reverse engineer the source of new models, I’m wondering what will stop companies from being fully transparent (besides ethics of course)