人工智能生成内容的著作权法之辩

Debate on Copyright Law of Artificial Intelligence Generated Content

根据输入指令与输出内容的比例关系,可将人工智能生成内容依次分为孪生内容、伴生内容和原生内容。人工智能生成内容主要涉及三个著作权法问题。一是机器阅读的法律性质问题,法定许可说能更好地兼顾著作权专有属性与公有属性,实现权利行使与限制的动态平衡。二是生成内容的可著作权性问题,需要结合内容类型并考虑输入指令的潜在影响。同比例复制的孪生内容不属于作品,差异化表达的伴生内容属于演绎作品,创造性表现的原生内容属于原创作品。三是生成内容的著作权归属问题,除另有协议外,需要结合内容类型并考虑所有者、使用者的贡献程度。孪生内容因不属于作品,无人享有著作权;所有者、使用者对伴生内容的形成均有重大贡献,基于概括合作行为与意识,所有者、使用者共同享有著作权;只有所有者对原生内容的形成具有重大贡献,所有者单独享有著作权。

According to the proportional relationship between input instruction and output content,artificial intelligence generated content can be divided into twin content,companion content and original content in turn.Artificial intelligence generated content mainly involves three copyright law issues.First,the legal nature of machine reading.The theory of legal license can better balance the exclusive and public attributes of copyright and realize the dynamic balance between the exercise and restriction of rights.Second,the copyright of artificial intelligence generated content needs to be combined with the content type and the potential influence of input instructions,and the twin content copied in the same proportion does not belong to the work;The accompanying content of differentiated expression belongs to deductive works;The original content of creative expression belongs to original works.The third is the copyright ownership of the generated content.Unless otherwise agreed,it is necessary to combine the content type and consider the contribution of the owner and the user.Because the twin content does not belong to the work,no one enjoys the copyright;Both owners and users have made great contributions to the formation of associated content.Based on summarizing cooperative behavior and consciousness,owners and users enjoy copyright together.Only the owner has made a significant contribution to the formation of the original content,so the owner enjoys the copyright alone.