Research On The Legal Regulation Of Machine Learning Copyright

The development of machine learning is highly valued by the state,and its role in promoting social welfare and improving total social knowledge in the copyright field has gradually emerged.Machine learning is a key technology to realize artificial intelligence based on massive data,and the issue of its copyright legal regulation has attracted greater attention in the academic community.The use of works for creation should be legally defined through the copyright limitation and exception system,however,China did not legally regulate the use of works by machine learning when the Copyright Law was amended in 2020,and the legal boundary of the copyright of machine learning cannot be correctly defined with the existing system,and machine learning is easily caught in the storm of infringement,leading to the imbalance of the interest balancing mechanism.Therefore,in order to reasonably balance the interests of individuals and the public interests of society,there is an urgent need to regulate machine learning by copyright law.In view of the huge amount of data in the works involved in machine learning,it is difficult to realize the authorized license in practice,and academics advocate the adoption of a fair use system or a statutory license use system to regulate it.The study takes the principle of machine learning as the entry point,and uses literature analysis,case analysis,value analysis and comparative analysis to conduct research on the legal regulation of machine learning copyright in order to solve the problem of ambiguous legal adjustment.Firstly,it explains the definition and principle of machine learning,clarifies that machine learning has the same dichotomy of thought and expression as human learning,and should not be directly recognized as infringement,and makes legal characterization of machine learning creative acts to provide the theoretical basis for the later research.At the same time,the copyright risk and conflict of interest of machine learning are analyzed in detail,the urgency of legal regulation is explained and a preliminary conclusion is drawn,that is,there is a certain possibility of infringement,but the acts need to be judged in conjunction with the results of expression rather than being excluded from the copyright restriction system.On this basis,the analysis draws on the overseas experience,compares and studies the current operation of the fair use system and the statutory license use system in China,and analyzes the justification of the regulation by the fair use system and the defects and dilemmas of the current regulation by the statutory license system in light of the legislation and judicial practice.Through an in-depth analysis of the principles of machine learning,the fair use system,the statutory license use system,the study argues that it is not reasonable to directly identify the use of works by machine learning as infringing use,and suggests that the fair use system should be adopted for regulation at present,and explains the justification reasons.It also suggests the normative design of the fair use system,creating new fair use exceptions through administrative regulations,while making the transition from the adoption of the reasonable use system to the statutory permitted use system.Further,suggestions are made to reshape the rules of the fair use system,i.e.,to break the restrictions of applicable subjects,set the prerequisites for lawful access to data,and introduce judicial interpretations to reshape the fair use system,in order to solve the problem of ambiguity in the legal regulation of machine learning copyright under the premise of balancing interests.