在科技日新月异的今天，人工智能发展十分迅速，目前已经发展到生成式人工智能阶段。越来越多的人利用生成式人工智能进行创作，当然科技进步的同时也引发了许多著作权侵权问题。本研究通过分析生成式人工智能的创作过程，可以发现涉及的著作权侵权问题主要包括三种情况：第一，生成式人工智能输入数据侵犯著作权问题，即未经数据权利人的允许使用受保护数据的侵权行为。在该种情形下，会涉及数据提供者与生成式人工智能开发者的侵权责任。现有研究鲜少从《著作权法》角度对数据进行保护。本研究将数据区分为受著作权保护的数据、公有领域的数据以及数据集合，通过适用《著作权法》对有关数据进行保护是可行的路径。第二，生成式人工智能创作内容侵犯现有作品著作权，即生成式人工智能的创作内容与现有作品构成《著作权法》中的实质性相似，并侵犯现有作品著作权的行为。此种情形下，用户应当承担侵权责任，如果生成式人工智能开发者在收到权利人通知后未履行删除义务，则也应当承担侵权责任。生成式人工智能与现有作品的比较标准遵循“接触+实质性相似”的著作权侵权判断规则，并且要受到“思想/表达二分法”的限制。第三，侵犯生成式人工智能创作作品权利，即生成式人工智能创作内容符合一定要求的情况下构成作品，未经新作品权利人的同意使用该作品则构成侵权。生成式人工智能创作内容在具有人的先行创作行为以及符合《著作权法》中规定的独创性要求的情况下能够构成作品，并且权利主体为投入先行创作行为的自然人、法人或者其他组织。本文旨在从《著作权法》的角度探究生成式人工智能三种侵权情况，并逐一提出解决实践问题的办法。

In today's rapidly changing technology, artificial intelligence has developed rapidly and has now reached the stage of generative artificial intelligence. More and more people are using generative artificial intelligence for creation, and of course, technological progress has also led to many copyright infringement issues. This study analyzes the creative process of generative artificial intelligence and finds that the copyright protection issues involved mainly include three situations: First, the infringement of copyright by generative artificial intelligence data, that is, the infringement of using protected data without the permission of the data rights holder. In this situation, there will be infringement liability between data providers and developers of generative artificial intelligence. It is a feasible path to protect relevant data through the Copyright Law by distinguishing it into data protected by copyright and data in the public domain. Secondly, the copyright of existing works is infringed by the creative content generated by artificial intelligence. that is to say, the innovative material produced by generative artificial intelligence. is substantially similar to existing works in the Copyright Law and infringes upon the copyright of existing works. In this case, users should bear the liability for infringement. If the developer of generative artificial intelligence fails to fulfill the deletion obligation after receiving notification from the rights holder, they should also bear the liability for infringement. The comparison standard between generative artificial intelligence and existing works follows the copyright infringement judgment rule of "contact+substantive similarity", and is subject to the limitation of the "thought/expression dichotomy". Thirdly, infringement of the right to create works using generative artificial intelligence implies that if the content generated by generative AI fulfills the criteria of the Copyright Law. it constitutes a work, and if the work is used without the consent of the new work owner, it constitutes infringement. The content generated by generative artificial intelligence qualifies as a work when preceded by creative actions of an individual and fulfills the originality standards established under the Copyright Law. and the rights subject is the natural person, legal person, or other organization that participates in the prior creative behavior. Regardless of which of the three situations mentioned above, the tort liability of generative artificial intelligence still follows the general principle of fault liability attribution. In addition, the responsibility of developers of generative artificial intelligence applies the "notify delete" rule by analogy.