Translation by Katharin Tai and Rogier Creemers. Editing by Lorand Laskai and Graham Webster
TRANSLATION
Notice on the Issuance of the “Guiding Opinions on Strengthening Overall Governance of Internet Information Service Algorithms”
Document published by the CAC (2021) No. 7
All provincial, autonomous region, and municipal cybersecurity and informatization offices, Party Committee propaganda departments, education offices (education committees), science and technology offices (committees, bureaus), telecommunications management bureaus, public security offices (bureaus), culture and tourism offices (bureaus), market supervision bureaus (offices, committees), and radio and television bureaus; the Xinjiang Production and Construction Corps cybersecurity and informatization office, Party Committee propaganda department, education bureau, science and technology bureau, public security bureau, culture, sports, radio, television, and tourism bureaus, and market regulation bureau:
In order to strengthen the overall governance of Internet information service algorithms, and stimulate the healthy, orderly, and flourishing development of the sector, nine ministries and commissions, the Cyberspace Administration of China, the Central Propaganda Department, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security, the Ministry of Culture and Tourism, the State Administration of Market Regulation, and the National Radio and Television Administration, have formulated the “Guiding Opinions on Strengthening Overall Governance of Internet Information Service Algorithms.” They are hereby issued to you; please integrate them into practice and earnestly implement them.
Cyberspace Administration of China
Central Propaganda Department
Ministry of Education
Ministry of Science and Technology
Ministry of Industry and Information Technology
Ministry of Public Security
Ministry of Culture and Tourism
State Administration of Market Regulation
National Radio and Television Administration
Sept. 17, 2021
Guiding Opinions on Strengthening Overall Governance of Internet Information Service Algorithms
In recent years, Internet information service algorithms (hereinafter referred to as "algorithms") have played an important role in areas like accelerating the dissemination of Internet information, enrichiching the digital economy, and promoting social development. At the same time, the unreasonable use of algorithms also affects the regular order of information dissemination, markets, and society, and poses challenges to safeguarding ideological security, social fairness and justice, and the legitimate rights and interests of netizens. In order to thoroughly implement the policy plans of the Party Central Committee and the State Council; to manage, use, and develop the application of algorithms well; and to comprehensively improve overall network governance capabilities; the following Opinions are hereby put forward on strengthening the security governance of Internet information service algorithms.
I. General Requirements
(1) Guiding Ideology
Adhere to the guidance of Xi Jinping’s Thoughts on Socialism with Chinese Characteristics for a New Era, especially General Secretary Xi Jinping’s important thinking on a cyber superpower [网络强国]; thoroughly implement the spirit of the 19th National Congress of the Communist Party of China and the 2nd, 3rd, 4th, and 5th Plenary Sessions of the 19th Central Committee. Uphold positive energy as the general requirement, management as a hard principle, good usage [用得好] as the true skill. Guided by the safe, reliable, high-quality, and innovative development of algorithms, establish a sound algorithm security governance mechanism, build a comprehensive algorithm security supervision system, promote independent innovation in algorithms, encourage the healthy, orderly, and prosperous development of algorithms, and provide strong support for building a cyber superpower.
(2) Basic Principles
Persevere in the correct orientation; strengthen awareness of scientific and technological ethics, safety, and the bottom line; give full play to the utility of algorithms for disseminating positive energy; and create a cyberspace of clean and upright winds [风清气正]. Uphold the rule of law, strengthen the construction of laws and regulations, innovate models of technological supervision, crack down on violations of laws and regulations, and establish a sound multiparty algorithm security governance mechanism. Insist on risk prevention and control, promote graded and categorized [分级分类] security management for algorithms, effectively identify high-risk algorithms, and implement precise governance; adhere to rights protection, and guide algorithm application to be fair and equitable, transparent, and fully interpretable; fully protect the legitimate rights and interests of netizens; and persist in technological innovation, vigorously promote China's algorithm innovation research work, protect algorithm intellectual property rights, strengthen the deployment and promotion of self-developed algorithms, and enhance the core competitiveness of China's algorithms.
(3) Main Objectives
Over the next three years, gradually establish a comprehensive algorithm security governance structure with a robust governance mechanism, a refined supervisory system, and a standardized algorithm ecosystem.
- Complete governance mechanism. Formulate robust policies and statutes for the governance of Internet information service algorithm security, with clear rights and responsibilities for algorithm security governance, an efficiently operating governance structure, a governance mechanism with reliable laws, and multifaceted cooperation and multiparty participation taking shape.
- Refined supervisory system. Innovatively construct and form a multi-dimensional, integrated supervision system that includes algorithm security risk monitoring, algorithm security assessments, investigation into science and technology ethics, algorithm filing management, and the handling of algorithm-related violations of laws and regulations.
- Standardized algorithm ecosystem. Algorithms are to be oriented properly with abundant positive energy; the application of algorithms is to be fair and just, and disclosed and transparent; the development of algorithms is to be secure and controllable and independently innovated, and to effectively prevent hidden risks caused by the abuse of algorithms.
II. Completing Algorithm Security Governance Mechanisms
(1) Strengthen the standards for algorithm governance. Refine algorithm security governance policies and regulations; accelerate the formulation of algorithm management regulations; clarify algorithm management subjects, scope, requirements, legal responsibilities, etc.; improve algorithm security governance measures; and formulate supporting documents such as standards and guidelines.
(2) Optimize the algorithmic governance structure. Further clarify the rights, obligations, and responsibilities of government, enterprises, industry organizations, and netizens in algorithm security governance; scientifically and rationally lay out the organizational structure for governance; standardize operations, and connect them with each other; create a multiparty, common algorithm security governance situation of government supervision, enterprises fulfilling their obligations, industry self-discipline, and social supervision.
(3) Strengthen overall coordination and governance. The cybersecurity and informatization department, together with the propaganda, education, science and technology, industry and information technology, public security, culture and tourism, market supervision, radio and television, and other departments, are to establish a long-term mechanism for departmental coordination and joint action, perform supervisory responsibilities, and jointly develop algorithm security governance.
(4) Strengthen the main responsibilities of enterprises. Enterprises should establish a system responsible for algorithm security, complete algorithm security management organisation mechanisms, strengthen risk prevention and dealing with troubleshooting, and increase the capacity and level of algorithm security emergency handling. Enterprises should strengthen awareness of their responsibilities, and take primary responsibility for the outcomes produced in applying algorithms.
(5) Strengthen self-discipline of industry organizations. The Internet information service industry should strengthen industry self-discipline, actively develop the task of popularizing algorithm science and technology, gradually build algorithm security governance capabilities, attract ranks of expert talent, gather resource input from multiple areas, assume social responsibility for algorithm security governance, and provide strong support for algorithm security governance.
(6) Advocate for netizen oversight and participation. Encourage netizens at large to actively participate in algorithm security governance, and effectively strengthen information exchange and effective communication between the government, enterprises, industry organizations, and netizens. Have the government actively accept reports and complaints from netizens, and have enterprises conscientiously accept social supervision and provide timely feedback on the outcomes.
III. Building Supervision and Management Systems for Algorithm Security
(7) Effectively monitor algorithm security risks. Conduct daily monitoring work of the data use, application landscape, impact, effects, etc., of algorithms; perceive changes in online dissemination trends, market norms, netizens’ activities, etc., brought about by the use of algorithms; provide early warning about risks such as where the use of algorithms may engender non-standardized, unfair, and unjust outcomes; and discover security problems with the application of algorithms.
(8) Vigorously launch assessments of algorithm security. Organize and establish specialized technology assessment teams; deeply analyze algorithmic mechanisms; assess the flaws and vulnerabilities in algorithm design, deployment, use, and other application segments; research the ideological, social justice, moral, ethical, and other security risks engendered by the application of algorithms; and put forward focused response measures.
(9) Advance algorithm filing work in an orderly manner. Establish algorithm filing systems, comb through the basic situation of algorithm filing, complete algorithm grading and categorization systems, clarify the scope of algorithm filing, and conduct filing work in an orderly manner. Vigorously guide and assist filing, actively publish filing status, and accept social supervision.
(10) Continue to advance innovation in supervision and management models. Continue to research new technology development trends in the field of algorithms; advance the coordinated development of supervision and management models and algorithm technology; continuously perfect, upgrade, and innovate supervision methods and governance measures; and prevent backwardness in supervision and management models from leading to algorithm security risks.
(11) Strictly strike against activities that violate laws and regulations. Strive to resolve algorithm security problems to which netizens react strongly; strictly attack algorithm-related activities violating laws and regulations discovered during algorithm monitoring, assessment, filing, and other such work, as well as verified reports from netizens; and firmly uphold the security of Internet information service algorithms.
IV. Promoting the Standardized Development of the Algorithm Ecosystem
(12) Establish the correct orientation for algorithms. Carry forward the socialist core values view, uphold the correct political direction, public opinion orientation, and value orientation in the application of algorithms. Raise the accuracy and efficacy of positive energy dissemination, standardize information distribution activities and order, promote enterprises to strengthen the dissemination of positive energy with the help of algorithms, and guide the application of algorithms to strive upward and for good.
(13) Promote openness and transparency of algorithms. Standardize the conduct of enterprise algorithm application, protect the lawful rights and interests of netizens, uphold the principles of fairness and justice, and stimulate the openness and transparency of algorithms. Urge enterprises to disclose basic algorithm principles, optimization objectives, decision-making standards, and other such information in a timely, reasonable, and effective manner. Do a good job explaining the results of algorithms, open up complaints channels, eliminate social misgivings, and promote the healthy development of algorithms.
(14) Encourage algorithm innovation and development. Increase algorithm innovation capabilities, vigorously launch algorithm research and development work, support the profound integration of algorithms with all areas of society and the economy. Raise independent and controllable algorithm capabilities, strengthen the protection of intellectual property rights, popularize the use of self-learning algorithm products, and strengthen core competitiveness in algorithms.
(15) Prevent the risks of algorithm abuse. Protect the order of dissemination, markets, and society, in cyberspace; prevent exploitative uses of algorithms that interfere with societal public opinion, suppress competition, harm netizens’ rights and interests, and other such acts; and prevent hidden risks in areas such as ideology, economic development, and social management caused by the abuse of algorithms.
CHINESE-LANGUAGE ORIGINAL
关于印发《关于加强互联网信息服务算法综合治理的指导意见》的通知
国信办发文〔2021〕7号
各省、自治区、直辖市网信办、党委宣传部、教育厅（教委）、科技厅（委、局）、通信管理局、公安厅（局）、文化和旅游厅（局）、市场监管局（厅、委）、广播电视局，新疆生产建设兵团网信办、党委宣传部、教育局、科技局、公安局、文化体育广电和旅游局、市场监管局：
为加强互联网信息服务算法综合治理，促进行业健康有序繁荣发展，国家互联网信息办公室、中央宣传部、教育部、科学技术部、工业和信息化部、公安部、文化和旅游部、国家市场监督管理总局、国家广播电视总局等九部委制定了《关于加强互联网信息服务算法综合治理的指导意见》。现印发给你们，请结合实际，认真贯彻执行。
国家互联网信息办公室
中央宣传部
教育部
科学技术部
工业和信息化部
公安部
文化和旅游部
国家市场监督管理总局
国家广播电视总局
2021年9月17日
关于加强互联网信息服务算法综合治理的指导意见
近年来，互联网信息服务算法（以下简称“算法”）在加速互联网信息传播、繁荣数字经济、促进社会发展等方面发挥了重要作用。与此同时，算法不合理应用也影响了正常的传播秩序、市场秩序和社会秩序，给维护意识形态安全、社会公平公正和网民合法权益带来挑战。为深入贯彻落实党中央、国务院决策部署，管理好使用好发展好算法应用，全面提升网络综合治理能力，现就加强互联网信息服务算法安全治理提出如下意见。
一、总体要求
（一）指导思想
坚持以习近平新时代中国特色社会主义思想特别是习近平总书记关于网络强国的重要思想为指导，深入贯彻党的十九大和十九届二中、三中、四中、五中全会精神，坚持正能量是总要求、管得住是硬道理、用得好是真本事，以算法安全可信、高质量、创新性发展为导向，建立健全算法安全治理机制，构建完善算法安全监管体系，推进算法自主创新，促进算法健康、有序、繁荣发展，为建设网络强国提供有力支撑。
（二）基本原则
坚持正确导向，强化科技伦理意识、安全意识和底线思维，充分发挥算法服务正能量传播作用，营造风清气正的网络空间；坚持依法治理，加强法律法规建设，创新技术监管模式，打击违法违规行为，建立健全多方参与的算法安全治理机制；坚持风险防控，推进算法分级分类安全管理，有效识别高风险类算法，实施精准治理；坚持权益保障，引导算法应用公平公正、透明可释，充分保障网民合法权益；坚持技术创新，大力推进我国算法创新研究工作，保护算法知识产权，强化自研算法的部署和推广，提升我国算法的核心竞争力。
（三）主要目标
利用三年左右时间，逐步建立治理机制健全、监管体系完善、算法生态规范的算法安全综合治理格局。
——治理机制健全。制定完善互联网信息服务算法安全治理政策法规，算法安全治理的主体权责明确，治理结构高效运行，形成有法可依、多元协同、多方参与的治理机制。
——监管体系完善。创新性地构建形成算法安全风险监测、算法安全评估、科技伦理审查、算法备案管理和涉算法违法违规行为处置等多维一体的监管体系。
——算法生态规范。算法导向正确、正能量充沛，算法应用公平公正、公开透明，算法发展安全可控、自主创新，有效防范算法滥用带来的风险隐患。
二、健全算法安全治理机制
（一）加强算法治理规范。健全算法安全治理政策法规，加快制定算法管理规定，明确算法管理主体、管理范围、管理要求和法律责任等，完善算法安全治理措施，制定标准、指南等配套文件。
（二）优化算法治理结构。进一步明确政府、企业、行业组织和网民在算法安全治理中的权利、义务和责任，科学合理布局治理组织结构，规范运作、相互衔接，打造形成政府监管、企业履责、行业自律、社会监督的算法安全多元共治局面。
（三）强化统筹协同治理。网信部门会同宣传、教育、科技、工信、公安、文化和旅游、市场监管、广电等部门，建立部门协同联动长效机制，履行监管职责，共同开展算法安全治理工作。
（四）强化企业主体责任。企业应建立算法安全责任制度和科技伦理审查制度，健全算法安全管理组织机构，加强风险防控和隐患排查治理，提升应对算法安全突发事件的能力和水平。企业应强化责任意识，对算法应用产生的结果负主体责任。
（五）强化行业组织自律。互联网信息服务行业应当加强行业自律，积极开展算法科学技术普及工作，逐步组建算法安全治理力量，吸引专业人才队伍，汇聚多方资源投入，承担算法安全治理社会责任，为算法安全治理提供有力支撑。
（六）倡导网民监督参与。鼓励广大网民积极参与算法安全治理工作，切实加强政府、企业、行业组织和网民间的信息交流和有效沟通。政府积极受理网民举报投诉，企业自觉接受社会监督并及时做好结果反馈。
三、构建算法安全监管体系
（七）有效监测算法安全风险。对算法的数据使用、应用场景、影响效果等开展日常监测工作，感知算法应用带来的网络传播趋势、市场规则变化、网民行为等信息，预警算法应用可能产生的不规范、不公平、不公正等隐患，发现算法应用安全问题。
（八）积极开展算法安全评估。组织建立专业技术评估队伍，深入分析算法机制机理，评估算法设计、部署和使用等应用环节的缺陷和漏洞，研判算法应用产生的意识形态、社会公平、道德伦理等安全风险，提出针对性应对措施。
（九）有序推进算法备案工作。建立算法备案制度，梳理算法备案基本情况，健全算法分级分类体系，明确算法备案范围，有序开展备案工作。积极做好备案指导帮助，主动公布备案情况，接受社会监督。
（十）持续推进监管模式创新。持续研判算法领域技术发展新形势，推进监管模式与算法技术协同发展，不断完善、升级、创新监管的方式方法和治理举措，防范监管模式落后导致的算法安全风险。
（十一）严厉打击违法违规行为。着力解决网民反映强烈的算法安全问题，对算法监测、评估、备案等工作中发现的、以及网民举报并查实的涉算法违法违规行为，予以严厉打击，坚决维护互联网信息服务算法安全。
四、促进算法生态规范发展
（十二）树立算法正确导向。弘扬社会主义核心价值观，在算法应用中坚持正确政治方向、舆论导向、价值取向。提高正能量传播的精准性和有效性，规范信息分发行为和秩序，推动企业借助算法加强正能量传播，引导算法应用向上向善。
（十三）推动算法公开透明。规范企业算法应用行为，保护网民合理权益，秉持公平、公正原则，促进算法公开透明。督促企业及时、合理、有效地公开算法基本原理、优化目标、决策标准等信息，做好算法结果解释，畅通投诉通道，消除社会疑虑，推动算法健康发展。
（十四）鼓励算法创新发展。提升算法创新能力，积极开展算法研发工作，支持算法与社会、经济各领域深度结合。提高算法自主可控能力，加强知识产权保护，提高自研算法产品的推广和使用，增强算法核心竞争力。
（十五）防范算法滥用风险。维护网络空间传播秩序、市场秩序和社会秩序，防止利用算法干扰社会舆论、打压竞争对手、侵害网民权益等行为，防范算法滥用带来意识形态、经济发展和社会管理等方面的风险隐患。
