Research On Legal Regulation Of Algorithms

The rapid iteration and widespread application of algorithms have created a more efficient and convenient world.At the same time,it has also brought a series of legal and ethical risks and challenges.The first part of this article analyzes the insecure purposes,insecure results,robustness problem,black-box problem,algorithmic discrimination,and control ability of algorithms.Thus,the legal regulation of artificial intelligence algorithms is imperative.Currently,China’s legislation on algorithms is relatively rough,and it is basically regulated indirectly through laws on network security and data security.The direct regulations on algorithms in the Personal Information Protection Law and the E-commerce Law are relatively scattered.Moreover,the specialized legislation around algorithms is mainly concentrated at the level of the departmental regulation,with strong limitations on the legal effect and regulatory objects.The second part of this article systematically summarizes the existing algorithmic governance schemes and their shortcomings in China,and analyzes the legislative experiences of the European Union and the United States to provide reference and suggestions.Firstly,stand at the hierarchy of the law,make separate legislation on algorithms.Secondly,in the process of legislation,attention should be paid to the allocation of diversified stakeholders’ rights and responsibilities that cover the entire lifecycle of the algorithm.Thirdly,when setting responsibilities and obligations,it should be based on the algorithm classification and grading,and follow the proportionality principle to maximize the balance between protection,development,and innovation.Due to China’s current academic research on algorithmic governance is still relatively broad,mostly focusing on path planning and legal architecture system design at the macro level,the third part of this article pays special attention to the design of regulatory schemes at the executable level.The article refers to the existing experience of data classification and grading,and discusses how to construct an algorithm classification and grading system.From the perspective of algorithm classification,this article proposes four principles that classification should follow,and proposes various classification modes,such as classification based on the control subject of the algorithm or the crowd targeted by the algorithm.From the perspective of algorithmic grading,this article emphasizes the concept of taking risk as the core,and puts forward the principles of enforceability,rigor,objectivity,and timeliness.In addition,the evaluation levels are designed based on the influencing degree and influencing objects such as countries,societies,organizations,and individuals.Besides,the third part of this article also discusses the subsequent algorithm security evaluation and algorithm filing mechanism.This article aims to propose a specific scheme based on the classification and grading system for algorithmic regulation,filling the governance system with operable details.