规范算法推荐 发展科技法理
Return to the Library

Standardized Algorithm Recommendations: Developing Science and Technology Legal Theory

规范算法推荐 发展科技法理

This publication by the Cyberspace Administration of China explains the legal principles behind the recently published, first of its kind, algorithm regulations.


FacebookTwitterLinkedInEmailPrintCopy Link
Original text
PDF
English text
PDF
See an error? Drop us a line at
View the translated and original text side-by-side

With the increasingly widespread application of digital technology, algorithms are gradually playing an increasingly important role in social production and daily life. We are gradually entering a world where “algorithms are everywhere.” Algorithms are not only bridges that allow people to obtain useful information in the ever-expanding Internet information ocean, but also the keys to making data and hardware resources better serve their functions. They are of great significance in the construction of digital China. However, at the same time, with the continuous increase in social digitization and intelligence, many issues such as algorithm security, algorithm black boxes, algorithm discrimination, and information cocoons (信息茧房) have received a high level of attention worldwide. Countries are actively exploring how to better respond and solve these issues. On December 31, 2021, the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation jointly issued the Provisions on the Administration of Algorithm-generated Recommendations for Internet Information Services (hereinafter referred to as the Provisions) to be implemented from March 1, 2022. The Provisions aim to regulate the algorithm recommendation activities of Internet information services, safeguard national security and the public interest, protect the legitimate rights and interests of citizens, legal persons, and other organizations, and promote the healthy development of Internet information services. The publication and implementation of the Provisions is a positive response to algorithm recommendation, a prominent issue of widespread concern in China’s network governance field. It is a specific legal norm established for the recommendations of Internet information service algorithms, the first of its kind in the world. It is a useful supplement to the legal order system of an intelligent society and an important exploration for the in-depth integration and development of the construction of “Digital China” and “China under the Rule of Law.” The specific normative measures in the Regulations reflect China’s exploration, development, and implementation of science and technology (S&T) legal theory in S&T development and application. In particular, it embodies the following principles:

随着数字技术日益广泛的应用,算法逐渐在社会生产、生活中发挥着日益重要的作用。我们正逐渐迈入“算法泛在”的世界。算法既是使人们在日渐广袤的互联网信息海洋中获得有益信息的桥梁,也是使数据和硬件资源得以更加充分地发挥作用的关键所在,在数字中国建设中具有重要意义。但与此同时,随着社会数字化、智能化程度的不断提高,算法安全、算法黑箱、算法歧视、信息茧房等诸多问题在世界范围内得到了高度关注,各国都在积极探索如何更好地对此予以回应和解决。2021年12月31日,国家互联网信息办公室、工业和信息化部、公安部、国家市场监管总局联合发布《互联网信息服务算法推荐管理规定》(以下简称《规定》),于2022年3月1日起开始实施。《规定》旨在规范互联网信息服务算法推荐活动,维护国家安全和社会公共利益,保护公民、法人和其他组织的合法权益,促进互联网信息服务健康发展。《规定》的发布和实施,是对算法推荐这一当下我国网络治理领域具有突出性的、受到广泛关注问题的积极回应,是在世界范围内首次针对互联网信息服务算法推荐建立起来的具体法律规范,是智能社会的法治秩序体系的有益补充,是“数字中国”与“法治中国”建设深度融合发展的重要探索。《规定》中的具体规范措施,体现了我国在科技发展运用之中对于科技法理的探索、发展和落实,特别体现了如下原则:

I. The Principle of Combining Ethics with the Rule of Law

一、伦理与法治相结合原则

The Provisions adopt a rule-setting model that combines overall principles with ethical rules in specific fields and links S&T ethics with entity responsibility, combining ethics and the rule of law in a more organic and scientific manner. In recent years, many countries, regions, organizations, and enterprises have successively issued ethical guidelines for artificial intelligence (AI) and automated systems. These guidelines received great interest from China. For example, the Governance Principles for a New Generation of Artificial Intelligence: Developing Responsible Artificial Intelligence published by the National New Generation Artificial Intelligence Governance Specialist Committee achieved a high degree of consensus on the values of putting people first, promoting innovation, ensuring security, protecting privacy, and clarifying responsibilities. Article 4 of the Provisions stipulates that we “respect social morality and ethics, abide by business ethics and professional ethics, and follow the principles of fairness, openness, transparency, scientific rationality, and good faith.” Some important ethical principles in this field are absorbed as general principles to be followed by algorithm recommendation activities. At the same time, it also implemented specific targeted regulations for some key areas. For example, Article 8 stipulates that “it is prohibited to set up algorithm models that violate public order and good manners, such as inducing users to indulge in addiction or excessive consumption.” This two-level embedding of ethics in legal norms reflects that, in the stage of rapid technological development and changes and the ongoing exploration of the legal order for an intelligent society, we must rely to a certain extent on the idea of linking the rule of law with ethics. At the same time, in order to better guide enterprises to pay attention to S&T ethics and take corresponding measures, compared with the draft for comment released in August 2021, requirements for the establishment of management systems and technical measures such as “S&T ethics review” were specifically added in Article 7 of the Provisions as an important area for enterprises to implement their responsibilities as algorithm security entities.

《规定》中采用了整体原则与具体领域伦理规则相结合、科技伦理与主体责任相链接的规则设置模式,将伦理与法治更有机、科学地结合在一起。近几年来,众多国家、地区、组织、企业纷纷发布了人工智能或自动化系统的伦理准则,我国也予以了高度重视。例如国家新一代人工智能治理专业委员会发布的《新一代人工智能治理原则——发展负责任的人工智能》,在以人为本、促进创新、保障安全、保护隐私、明晰责任等价值观上取得了高度共识。《规定》第四条中规定了应“尊重社会公德和伦理,遵守商业道德和职业道德,遵循公正公平、公开透明、科学合理和诚实信用的原则”,吸纳了该领域的一些重要伦理原则,作为算法推荐活动应遵循的一般原则。同时,对于一些重点领域也进行了有针对性的具体规定。例如,第八条中规定“不得设置诱导用户沉迷或者高额消费等违背公序良俗的算法模型”。这种伦理在法律规范中的两层次嵌入,体现了在技术发展变化快速、智能社会法律秩序还在不断探索的阶段,在一定程度上依赖与伦理相链接的法治思路。同时,为更好地引导企业重视科技伦理并采取相应措施,相比2021年8月发布的征求意见稿,《规定》在第七条中更是专门增加了建立“科技伦理审查”等管理制度和技术措施的要求,作为企业落实算法安全主体责任的重要方面。

II. The Principle of Mutual Construction, Mutual Governance, and Mutual Enjoyment

二、社会共建共治共享原则

Ideas and theories such as “mutual construction, mutual governance, and mutual enjoyment” and building a “community of social governance” have pointed out an important development direction for social governance in the new era, and also laid a local theoretical foundation for the pluralistic mutual governance of algorithms. The Provisions focus on establishing specific norms for government supervision and management, the promotion of enterprise self-governance, the encouragement of industry self-discipline, the development of industry standards, and the facilitation of citizen rights protection. It is conducive to promoting the joint participation of diverse entities in algorithm governance, It is the specific implementation of the idea and basic theory of “mutual construction, mutual governance, and mutual enjoyment” in the field of algorithm governance.

“共建共治共享”、建设“社会治理共同体”等思想和理论,为新时代社会治理指出了重要发展方向,也奠定了算法多元共治的本土理论基础。《规定》中着重建立了政府监督管理、推动企业自治、鼓励行业自律、发展行业标准、便利公民维权等方面的具体规范,有利于推动多元主体共同参与算法治理,是“共建共治共享”的思想和基本理论在算法治理领域中的具体落地。

III. The Principle of Hierarchical and Classified Algorithm Governance

三、算法分级分类治理原则

Cyberspace is hierarchically divided and all-encompassing. Hierarchical and classified governance is a common principle in the rule of law for networks and information. This is embodied in the Cybersecurity Law, the Data Security Law, and the Personal Information Protection Law. Article 19 of the Provisions specifically stipulates that the cybersecurity and informatization departments, in conjunction with telecommunications, public security, market supervision, and other relevant departments, shall establish a hierarchical and classified security management system for algorithms, which is the concentrated embodiment of the principle of hierarchical and classified governance. In addition, currently, the principle of hierarchical and classified governance is mainly reflected in two aspects in the Provisions: First, for “algorithm recommendation services with public opinion attributes or social mobilization capabilities,” because they involve national security risks, in addition to the general rules in the Provisions, special obligations for “algorithm filing” and “self-assessment” are also established. Second, targeted regulations are established for some specific categories of services, such as the provision of services to minors (Article 18), the provision of services to the elderly (Article 19), the provision of work scheduling services to workers (Article 20), and the sale of goods or provision of services to consumers (Article 21). Hierarchical and classified governance avoids coarse-grained “one-size-fits-all” legislation and is conducive to enhancing the pertinence, scientific nature, and effectiveness of legal norms.

网络空间具有层次划分并且包罗万象,分级分类治理是网络与信息法治里面的普遍原则。这在《网络安全法》《数据安全法》《个人信息保护法》中均有具体体现。《规定》第十九条中专门规定了网信部门会同电信、公安、市场监管等有关部门建立算法分级分类安全管理制度,是分级分类治理原则的集中体现。此外,目前分级分类治理原则在《规定》中主要体现为两个方面:一是对于“具有舆论属性或者社会动员能力的算法推荐服务”,由于其涉及国家安全风险,因而除《规定》中的一般规则外,还特别设置了“算法备案”“自评估”的特别义务。二是面向一些特定类别的服务,分别进行了有针对性的规定,例如向未成年人提供服务(第十八条)、向老年人提供服务(第十九条)、向劳动者提供工作调度服务(第二十条)、向消费者销售商品或者提供服务(第二十一条)等。分级分类治理避免了“一刀切”式的粗犷立法,有利于增强法律规范的针对性、科学性、有效性。

IV. The Principle of Equal Emphasis on Security and Development

四、安全与发展并重原则

Resolving the conflicts between security and development, between order and vitality, is the difficulty and focus of the field of S&T rule of law. The formulation of the Provisions fully embodied the basic principle of giving equal emphasis to security and development. For example, the Provisions require that users be provided with a function to “select or delete” user labels, which reflects the comprehensive consideration of the interests of multiple parties and the scientific nature of the system. Another example is the establishment of the algorithm filing system, which embodies the characteristics of flexible governance and provides a variety of complex functions, including information collection functions, information publicity functions, and enterprise self-government incentive functions. Going forward, systems such as algorithm filing still need to be further refined. In the process of refinement, we must adhere to the principle of giving equal emphasis to security and development. China has some world-leading advantages in the new round of industrial revolution. To properly leverage these advantages and ensure the healthy development of the digital economy, it is urgently necessary to explore an effective path for algorithm governance. The promulgation and implementation of the Provisions not only provide a systematic governance path for algorithm issues that arise in practice, but also implement the legal principles summarized and developed in the practice of S&T rule of law in China. It has also made useful innovations and explorations in the construction of specific norms, providing an important foundation for the construction of a more comprehensive and deeper S&T legal system for the future.

如何解决好安全与发展、秩序与活力等目标之间的冲突,是科技法治领域的难点与重点。《规定》制定过程中充分体现了安全与发展并重的基本原则。例如,《规定》向用户提供“选择或者删除”用户标签功能的要求,反映了对多方利益和制度科学性的综合考量。再如,算法备案制度的建立,体现出柔性治理的特点,承载了多元化的复合型功能,包括信息收集功能、信息公示功能、企业自治激励功能等。未来算法备案等制度仍需进一步细化,在细化过程中也仍然需要坚持安全与发展并重的原则。我国在新一轮产业革命中具有一些领先世界的优势,要发挥好这种优势,保障数字经济的健康发展,亟需探索一条算法治理的有效路径。《规定》的出台和实施,既为在实践中出现的算法问题提供了系统的治理路径,又贯彻落实了在我国科技法治实践中总结和发展的法理原则,更在具体规范构建上进行了有益的创新和探索,为未来更加综合、深化的科技法治体系构建提供了重要基础。

To top

Cite This Page

"Standardized Algorithm Recommendations: Developing Science and Technology Legal Theory [规范算法推荐 发展科技法理]", CSIS Interpret: China, original work published in Cyberspace Administration of China (CAC) [国家互联网信息办公室], March 3, 2022

FacebookTwitterLinkedInEmailPrintCopy Link