人工智能文生视频大模型的作品风险、著作权归属及有效治理

Work Risks,Copyright Ownership and Effective Governance for Artificial Intelligence Text-to-video Large Model

以Sora为代表的文生视频大模型的出现,赋予作品风险、著作权归属和著作权风险治理新的讨论意蕴。笔者认为文生视频大模型作为人类精神产品创作的工具,其生成的内容构成作品。基于文生视频大模型的技术特点,其引发的作品风险不能一概而论,应基于对作品的不同使用行为加以识别和判断。在著作权归属领域,应坚持视听作品的私权属性,在秉持以意思自治约定归属优先的同时,在个案中结合实质贡献、投资激励和利益平衡原则确定著作权归属。在著作权风险治理领域,应通过著作权规则完善、著作权审判指导、著作权执法监管和大模型行业自治,共同作用于文生视频大模型的合规、高效与可持续发展。

The text-to-video large model represented by Sora has brought about new discussion on work risks,copyright ownership and copyright risk governance.Text-to-video large model is a tool for human intellectual products creation,and the content generated by it constitutes its work.Based on the technical characteristics of text-to-video large model,the risks of work should be identified and judged based on different usages of works rather than being generalized.In thefield of copyright ownership,we should adhere to the private property character of audiovisual works,take the party autonomy as the priority principle,and combine the substantial contribution,investment incentive and interest balance principles when distributing copyright ownership in cases.In thefield of copyright risk governance,the copyright law improvement,copyright trial guidance,copyright law enforcement supervision and large model trade autonomy should work together to the compliance,efficiency and sustainable development of the text-to-video large model.