Provisions on the Administration of Algorithmic Recommendations in Internet Information Services of the PRC — Full English Translation (2022)

Issued jointly by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation on December 31, 2021

Effective: March 1, 2022

Table of Contents

Chapter I — General Provisions

Article 1 — These Provisions are formulated in accordance with the Cybersecurity Law of the People’s Republic of China, the Data Security Law of the People’s Republic of China, the Personal Information Protection Law of the People’s Republic of China, the Internet Information Service Administrative Measures, and other relevant laws and administrative regulations, for the purposes of regulating the recommendation of algorithms in internet information services, promoting the healthy and orderly development of algorithm application services, safeguarding national security and the public interest, and protecting the lawful rights and interests of citizens, legal persons, and other organizations.

Article 2 — These Provisions apply to the use of algorithmic recommendation technologies to provide internet information services (hereinafter referred to as “algorithmic recommendation service providers”) within the territory of the People’s Republic of China. Where laws or administrative regulations provide otherwise, such provisions shall prevail.

For the purposes of the preceding paragraph, “algorithmic recommendation technologies” means technologies that use algorithms such as personalized push, search filtering, sorting and selection, retrieval and filtering, dispatching decisions, and other algorithmic technologies to provide information to users.

Article 3 — The national cyberspace administration authority shall be responsible for the overall coordination of the supervision and administration of algorithmic recommendation services nationwide. The departments for telecommunications, public security, market regulation, and other relevant departments shall, in accordance with their respective duties, be responsible for the supervision and administration of algorithmic recommendation services.

Article 4 — Algorithmic recommendation service providers shall adhere to the socialist core values, comply with laws and regulations, respect social morality and public order, and uphold a positive and healthy cyberspace. They shall not use algorithmic recommendation services to engage in activities prohibited by laws and administrative regulations, including activities that endanger national security and the public interest, disrupt the economic and social order, or infringe upon the lawful rights and interests of others.

Article 5 — Algorithmic recommendation service providers shall pursue fairness and justice, standardize the allocation of resources, promote the diversity of information dissemination, and protect the rights and interests of online users.

Article 6 — Algorithmic recommendation service providers shall adhere to the principles of openness and transparency, optimize the design of their algorithmic mechanisms, and provide users with convenient options for refusing algorithmic recommendations.

Chapter II — Information Service Standards

Article 7 — Algorithmic recommendation service providers shall implement the main responsibility for information content management, establish and improve management systems and technical measures for information content review, such as feature databases for identifying illegal and undesirable information, user registration and account management, review and editing of information content, and model training optimization, to prevent the dissemination of illegal and undesirable information.

Where illegal information is discovered, algorithm recommendation service providers shall immediately cease the transmission of such information, take measures to eliminate it, prevent its further dissemination, retain relevant records, and report to the relevant cyberspace administration authority and other relevant departments. Where undesirable information is discovered, algorithm recommendation service providers shall handle it in accordance with laws, administrative regulations, and departmental rules.

Algorithmic recommendation service providers shall strengthen the management of user accounts in respect of their algorithmic recommendation service sections, pages, and other content, and shall not present illegal and undesirable information.

Article 8 — Algorithmic recommendation service providers shall periodically review, evaluate, and verify the mechanisms, principles, models, and data and application results of their algorithms, and shall not set up algorithm models that induce users to become addicted or to spend excessive amounts of time or money. They shall not use algorithms to circumvent supervision and administration, or to engage in monopolistic conduct or unfair competition in violation of laws and regulations.

Article 9 — Algorithmic recommendation service providers shall strengthen positive dissemination of information. Where algorithmic recommendation service providers provide internet news information services, they shall obtain a license for internet news information services in accordance with the law, regulate the use of algorithms to conduct news information dissemination activities, and shall not generate or disseminate false news information.

Article 10 — Algorithmic recommendation service providers shall not use algorithms to create false accounts, engage in account trafficking, create false followers, or engage in click farming. They shall not use algorithms to manipulate rankings, search results, public opinion, or to interfere with the presentation of information. They shall not use algorithms to implement actions that influence online public opinion or circumvent supervision and administration.

Article 11 — Algorithmic recommendation service providers shall not use algorithms to provide users with illegal information or undesirable information. Where the use of algorithm recommendation services is discovered or suspected of being used to transmit illegal information, measures shall be taken in accordance with Article 7 of these Provisions.

Article 12 — Algorithmic recommendation service providers shall encourage the use of algorithms to disseminate positive energy and disseminate socialist core values. They shall use algorithms to recommend high-quality information and enhance the positive guidance of content.

Article 13 — Where algorithmic recommendation service providers provide internet information services to minors, they shall facilitate access to information beneficial to the healthy development of minors in accordance with the law, and shall not push information harmful to the physical and mental health of minors to minors, nor shall they use algorithms to provide services that may induce minors to become addicted. Algorithmic recommendation service providers shall develop corresponding management systems and protective measures.

Article 14 — Where algorithmic recommendation service providers provide internet information services to the elderly, they shall facilitate access to information beneficial to the elderly in accordance with the law. Algorithmic recommendation service providers shall take measures such as optimizing search results, providing barrier-free access, and preventing online fraud, to protect the lawful rights and interests of the elderly.

Article 15 — Algorithmic recommendation service providers shall not use algorithms to implement unreasonable differential treatment, or to damage the lawful rights and interests of users, such as through discriminatory pricing. Where trading conditions are used to implement differential treatment, the provider shall comply with the provisions of laws and administrative regulations, and shall not use algorithms to engage in monopolistic conduct or unfair competition.

Chapter III — User Rights Protection

Article 16 — Algorithmic recommendation service providers shall inform users in a prominent manner of the circumstances of the algorithmic recommendation services provided to them, and shall make public the basic principles, purposes, intentions, and operational mechanisms of the algorithm recommendation services in an appropriate manner.

Article 17 — Algorithmic recommendation service providers shall provide users with the option not to have their personal characteristics targeted by the algorithm, or shall provide users with convenient options for refusing algorithmic recommendation services. Where a user chooses to refuse algorithmic recommendation services, the algorithm recommendation service provider shall immediately cease providing the relevant services.

Algorithmic recommendation service providers shall provide users with functions or options for selecting or deleting user tags used for the algorithmic recommendation services targeting their personal characteristics.

Algorithmic recommendation service providers shall apply such user tag management rules, and provide users with convenient options for turning off algorithmic recommendation services.

Article 18 — Algorithmic recommendation service providers shall, in accordance with the law, inform users of their rights to access, correct, delete, and port their personal information, and provide corresponding operational functions. Algorithmic recommendation service providers shall not use algorithmic means to obstruct users from exercising their rights in accordance with the law.

Article 19 — Where algorithmic recommendation service providers provide users with services that have a significant impact on their major interests, such as employment placement, order dispatch, and credit scoring, they shall explain the relevant circumstances and rules of the algorithm to users in an appropriate manner.

Article 20 — Where algorithmic recommendation service providers provide internet information services to workers through algorithms such as employment placement and work dispatching, they shall establish and perfect platform rules and algorithm models that protect the lawful rights and interests of workers such as fair treatment, appropriate remuneration, rest and leave, and occupational safety and health.

Article 21 — Where algorithmic recommendation service providers provide internet information services to consumers, they shall protect consumers’ rights to be informed, to make their own choices, to fair trade, and other lawful rights and interests. They shall not use algorithms to impose unreasonable differential treatment on consumers in terms of trading conditions such as transaction prices, or to implement conduct that deceives or misleads consumers.

Chapter IV — Supervision and Administration

Article 22 — Algorithmic recommendation service providers with public opinion attributes or social mobilization capabilities shall undergo a security assessment in accordance with the relevant provisions of the state when they launch services, apply new technologies, modify the form of provision of services, or otherwise change their algorithmic recommendation services.

Article 23 — Algorithmic recommendation service providers shall, in accordance with the law, retain records of online logs such as user registration information, user tag models, algorithmic model training and recommendation logs, and cooperate with the relevant authorities in conducting security assessments, supervision, and inspections.

Article 24 — Algorithmic recommendation service providers with public opinion attributes or social mobilization capabilities shall, within 10 working days of providing services, complete the filing of their algorithmic recommendation service provider name, service form, application field, algorithm type, algorithm self-assessment report, content to be publicized, and other information through the internet information service algorithm filing system. Where the information filed is changed, the change shall be handled within 10 working days of the change. Where the algorithm recommendation service is terminated, the cancellation of filing shall be handled within 20 working days of the termination, and appropriate arrangements shall be made.

Where the algorithm filing information is complete, the cyberspace administration authority shall make the relevant information public within 30 working days. Algorithmic recommendation service providers shall publicize the algorithm filing number in a prominent position.

Article 25 — The cyberspace administration authority shall, in conjunction with the relevant departments, formulate rules and standards for the classified and graded security management of algorithm recommendation services and establish and improve a system for the supervision and administration of algorithm recommendation services. The relevant authorities shall perform their duties of supervision and administration in accordance with the law.

Article 26 — Where any organization or individual discovers any violation of laws, regulations, or these Provisions in algorithmic recommendation services, they may file a complaint or report with the cyberspace administration authority and other relevant departments. The relevant departments shall handle the complaint or report in a timely manner in accordance with the law.

Chapter V — Legal Liability

Article 27 — Where algorithmic recommendation service providers violate the provisions of Articles 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, or 23 of these Provisions, the cyberspace administration authority and other relevant departments shall impose penalties in accordance with the Cybersecurity Law of the People’s Republic of China, the Data Security Law of the People’s Republic of China, the Personal Information Protection Law of the People’s Republic of China, the Internet Information Service Administrative Measures, and other laws, administrative regulations, and departmental rules.

Article 28 — Where algorithmic recommendation service providers violate the provisions of Article 24 of these Provisions by failing to complete the filing of their algorithm recommendation services, the cyberspace administration authority and other relevant departments shall order them to make corrections within a prescribed time limit. Where they refuse to make corrections or the circumstances are serious, they shall be ordered to suspend the provision of information services, and a fine of not less than RMB 10,000 but not more than RMB 100,000 shall be imposed.

Article 29 — Where algorithmic recommendation service providers with public opinion attributes or social mobilization capabilities violate the provisions of these Provisions, the cyberspace administration authority and other relevant departments shall impose penalties in accordance with the provisions of laws, administrative regulations, and departmental rules. Where serious consequences are caused, the authorities may, in accordance with the law, order the suspension of information updates on the relevant business or the revocation of the relevant business licenses or permits.

Article 30 — Where algorithmic recommendation service providers violate the provisions of these Provisions, in addition to being subject to administrative penalties in accordance with laws and administrative regulations, the cyberspace administration authority and other relevant departments may impose measures such as issuing a warning, conducting regulatory talks, ordering the publication of an apology, or ordering rectification, and where serious circumstances exist, may order the suspension of information updates on the relevant business.

Chapter VI — Supplementary Provisions

Article 31 — For the purposes of these Provisions, algorithm recommendation service providers with “public opinion attributes or social mobilization capabilities” means algorithm recommendation service providers that provide internet information services such as internet news information services, services that have the function of public opinion expression, or services that have the capability to mobilize the public to engage in specific activities.

Article 32 — These Provisions shall take effect on March 1, 2022.

Free PDF download of the complete article.