人工智能生成内容著作权规制的全球趋向与本土路径

How Copyright Law Regulate AI Generated Content:Global Trends and Local Approaches

生成式人工智能技术的普及对著作权制度的影响主要体现在两个方面:一是输出端生成内容的可版权性与权利归属,二是输入端基于机器学习大规模使用他人作品的合法性认定。对于前者而言,在普遍坚持和认同自然人参与创作为可版权性前提的同时,全球各国对权利归属仍存在使用者和设计者的认知差别。对于后者而言,人工智能技术领先型和追赶型国家的制度选择存在较大差异:技术领先型国家更多借助判例法传统,等待传统版权产业与人工智能产业双方的充分博弈,延续以往平衡应对新旧产业冲突的路径,期待双方在充分表达利益的基础上实现产业合作途径的创新;技术追赶型国家的立法选择则更偏向于为人工智能产业提供发展空间。鉴于现阶段的技术水平和产业地位,我国有必要选择将技术追赶型国家的制度经验融入本土“三步检验法”,在机器学习的合理使用适用上破除“非营利性”和“适度性”局限。

The spread of generative AI technology impacts copyright system in two key areas:first at the output end,the copyrightability and ownership of AI-generated content,second at the input end,the legal boundaries of large-scale use of others'works based on machine learning.On the first issue,there are both consensus and disputes among the representative countries in the world.While universal adherence and recognition that human involvement is a prerequisite for copyright,there are still cognitive disputes between users and designers over ownership.On the second issue,there is a big difference between AI technology-leading and catching-up countries in their institutional choice.Technology-leading countries rely more on case law tradition,waiting for the game between the traditional copyright industry and the emerging AI industry,to continue the approach of dealing with the Internet industry in the past,and to expect the industrial cooperation channels and development space based on full expression of their interests and positions.By contrast,technology catching-up countries are more inclined to legislative approach to support AI industry.Given China's current technological level and industrial status,it is necessary to integrate institutional experience of technology catching-up countries into the interpretation and adjustment of the local"three-step test",and breaking the limits of"nonprofit"and"moderateness"in machine learning.