随着人工智能技术的持续演进与广泛应用，其通过神经网络学习机制所生成的文字、音乐、绘画及图形等内容，在外在表现上已与人类作品日趋接近，由此对传统著作权制度提出了新的挑战。尽管此类生成物在形式上呈现出作品样态，但仍未脱离人类的智力投入与技术支配，本质上仍应被理解为人类智力劳动的延伸性成果，归根到底仍应落实于人类主体。面对在人工智能生成物著作权归属中存在法律定性、权利归属和署名混乱的困境。提出构建明确考量因素、以归属人工智能使用者为原则和完善署名规则的著作权归属特有制度。

As artificial intelligence advances and proliferates, outputs generated via neural networks—text, music, and visual works—increasingly resemble human creations, challenging traditional copyright frameworks. Although formally work-like, such outputs depend on human intellectual input and technological guidance, and should be regarded as extensions of human labor. To address ambiguities in legal characterization, ownership, and authorship of AI-generated content, this paper proposes a sui generis copyright framework emphasizing clear criteria, user centered ownership, and refined attribution rules.