AI companies are sourcing vast amounts of text, images, and data to train models, including rare books and copyrighted material often obtained without creator consent. This practice raises questions about intellectual property rights, artist compensation, and whether AI development should operate under different rules than traditional media use.