Abstract: "Face-changing technology" is an intelligent video processing technology that uses the deep learning capabilities of the Internet to form a confrontational generation network model. Compared with traditional technology, this technology has the characteristics of high degree of authenticity, universal adaptability and rapid evolution. In the field of copyright law, due to the autonomous learning nature of this technology, the attributes of the “face-changing video” produced by itself have yet to be determined. The abuse of this technology not only infringes on the personal rights of the original work copyright owner and performer, but also harms the public interest. Article 22 of the current Copyright Law cannot fully explain the fair use of this technology. In the next stage, on the basis of existing technology and legal regulations, we must clarify the copyright law attributes of the technology product, build a "producer-platform-audience" pre-regulation system, clarify producers’ declaration obligations, strengthen platform review obligations, and strengthen The ability of people to identify information.

Key words: face-changing technology, artificial intelligence, copyright law, fair use, regulatory means

摘要： “换脸技术”是利用人工智能的深度学习能力，通过对抗生成网络模型形成的智能视频处理技术。与传统技术相比，“换脸技术”具有高度真实性、普遍适应性与快速演进性的特点。在著作权法领域中，其生产的“换脸视频”是否具有独创性以及是否需要法律保护存在较大争议。同时，该技术的滥用将侵犯原作品著作权人、原作品表演者的合法权利，而现行著作权法第22条难以解释其合理使用性。下一阶段，要在现有技术与法律规制基础上，明确该技术产物的著作权法保护手段，并构建“制作者-平台-受众”的事前规制体系，明确制作者申明义务，强化平台审查义务，增强民众识别信息的能力。

关键词: 换脸技术, 人工智能, 著作权法, 合理使用, 规制手段