The article explores the broader implications of policy choices regarding the training and ownership of generative AI (AI) outputs, focusing on two critical questions: how to compensate data providers for training models and who should own the outputs of AI. It discusses the challenges of proving copyright infringement when using pirated or illegally obtained data and the concept of fair use. The article also examines the legal status of "computer-generated" works, where the author is deemed to be the person who arranged the necessary arrangements for creation. It highlights the potential impact of AI on various sectors, including the arts and knowledge economy, and suggests that regulatory steps such as transparency in model development and deployment are crucial. The article concludes by emphasizing the need for a balanced approach to protect human creativity while encouraging innovation and the development of new tools enhanced by AI.The article explores the broader implications of policy choices regarding the training and ownership of generative AI (AI) outputs, focusing on two critical questions: how to compensate data providers for training models and who should own the outputs of AI. It discusses the challenges of proving copyright infringement when using pirated or illegally obtained data and the concept of fair use. The article also examines the legal status of "computer-generated" works, where the author is deemed to be the person who arranged the necessary arrangements for creation. It highlights the potential impact of AI on various sectors, including the arts and knowledge economy, and suggests that regulatory steps such as transparency in model development and deployment are crucial. The article concludes by emphasizing the need for a balanced approach to protect human creativity while encouraging innovation and the development of new tools enhanced by AI.