This article investigates the intersection of copyright law and artificial intelligence (AI) training in the United States and European Union, focusing on the legal implications of using copyrighted materials for AI model development. Through an analysis of notable controversies surrounding AI platforms like ChatGPT and Google's Gemini, the paper examines the principles of fair use and transformative use in the U.S., as well as the EU's text and data mining exceptions. The study highlights the conceptual inconsistencies in applying copyright laws to AI training, contrasting it with traditional human learning processes. Key findings suggest that while U.S. courts may recognize transformative use as a defense for AI training, the commercial nature of such use could lead to infringement rulings. In the EU, recent directives have established specific exceptions for text and data mining, facilitating AI training while allowing rights holders to assert control over their works. The article concludes that existing legal frameworks are struggling to keep pace with rapid AI advancements, indicating a need for clearer guidelines and potential reforms to address the unique challenges posed by AI technologies in copyright law.