Artificial intelligence (AI) systems encompass various types, among which generative AI is a prominent category. Generative AI employs machine learning techniques to analyze input data, enabling it to reconstruct or create texts, images, audio-visual content, and even 3D models based on known data through algorithmic processes. From the training of AI systems to the generation of content, Generative AI is intricately connected to copyright law. First, the data inputted into AI systems may constitute copyright-protected subject matter, raising concerns about whether the AI training process infringes copyright. Second, whether the content generated by generative AI can be afforded copyright protection akin to human-created works is a fundamental issue warranting in-depth exploration under copyright law. In light of these considerations, this paper aims to examine the copyright issues arising from generative AI systems. The discussion will focus on two key questions: whether AI generated content qualifies as a copyrightable subject matter and whether the training process of generative AI involves copyright infringement. The research adopts a methodology combining literature review and comparative legal analysis. By analyzing regulatory frameworks and policy approaches in the United States and the European Union, this study seeks to elucidate the principles underlying copyright disputes related to generative AI systems. Concurrently, it reflects on the current legal provisions, practices, and scholarly perspectives in the domestic context, with the goal of proposing reference points for addressing copyright issues associated with generative AI in the future.