AI合成影视作品中肖像权可识别性的认定规则研究

Research on the Identification Rules of Identifiability of Portrait Rights in AI-Generated Film and Television Works

摘要: 人工智能时代的到来，催生了AI合成经典影视作品这样的新型创作形式。短视频内容创作者借助AI换脸、深度合成等技术，对影视作品中的角色形象进行替换，甚至是恶搞。这种行为引发了不少关于肖像权保护的争议，其中最为核心的就是如何认定它是否侵犯了演员的肖像权？传统的可识别性认定规则，在AI合成的场景下，面临着识别对象、识别主体、识别方式三重认定困境。通过对现行规则的规范分析，在认定对象上应从单一的面部特征扩展为包括动态特征、角色标志性元素在内的要素集合；在认定主体上，确立以一般社会公众标准为基础、技术识别能力为补充的双重标准协同机制；在认定方式上，引入合成可识别性概念，从真实对应到合成指向。这套规则体系，既回应了AI技术对传统法律带来的冲击，也为司法实践提供了更具操作性的认定标准，在保护演员肖像权和保障创作自由之间寻求平衡。

Abstract: The advent of the AI era has given rise to a new form of creation: AI-modified classic film and television works. Short video content creators use AI face-swapping and deep synthesis technologies to replace or even parody the character images in film and television works. This behavior has sparked many disputes over the protection of portrayal rights, with the core issue being how to determine whether it infringes upon the portrayal rights of actors. The traditional recognition rules for identifiability face three major challenges in the context of AI modification: the identification object, the identification subject, and the identification method. Through an analysis of the current rules, this paper proposes that the identification object should be expanded from a single facial feature to a set of elements including dynamic features and role-specific elements; the identification subject should establish a dual standard coordination mechanism based on the general public standard and supplemented by technical identification capabilities; and the identification method should introduce the concept of synthetic identifiability, shifting from real correspondence to synthetic indication. This rule system not only responds to the impact of AI technology on traditional law but also provides more operational recognition standards for judicial practice, seeking a balance between protecting actors’ portrayal rights and safeguarding creative freedom.