Risks and Solutions of Copyright Infringement in Artificial Intelligence-generated Creations
[Purposes]This paper aims to explore the conflicts and coordination between the application of Artificial Intelligence(AI)deep learning and copyright protection,in order to balance technological in-novation and rights protection,and propose reasonable copyright law countermeasures.[Methods]By ana-lyzing the current application status of AI deep learning and combining relevant provisions of copyright law,the mechanism of copyright infringement risk formation in the AI creation process is studied.Taking examples such as Baidu's Wenxin Yiyan,the three stages of AI creation are studied in detail.[Findings]It is believed that the main risks come from the data input and output stages.In the data input stage of AI creation,the collection and processing of existing works may involve infringement issues.In the output stage,the copyright ownership of AI-generated works is complex,involving the originality input of AI models and humans.In addition,different data sources also bring different infringement risks.At the same time,within the framework of existing systems,it is difficult to avoid illegal use of digital works in AI creation.[Conclusions]It is suggested to modify the fair use terms to include AI creation in the fair use framework,thereby legalizing the gray area of the AI industry.
NETL
NSTL
万方数据
