Algorithm Recommendation Regulations

第一条 Legislative basis and scope

Establishes the regulation's purpose: governing recommendation algorithm activities in online information services to protect national security, public interest, and individual rights while promoting healthy development of internet services. The regulation applies within China's territory.

第二条 Applicable technology definition

Defines recommendation algorithm technology as using synthetic generation, personalized delivery, ranking and filtering, search filtering, or scheduling and decision-making algorithms to deliver information to users.

第三条 Regulatory authority

National and local internet administration departments coordinate governance and oversight of recommendation services. Other relevant departments at national and local levels supervise these services according to their respective responsibilities.

第四条 Service provision principles

Providers of recommendation services must comply with laws and regulations, respect social ethics, uphold business and professional integrity, and adhere to principles of fairness, transparency, rationality, and good faith.

第五条 Industry self-regulation

Encourages industry organizations to establish and strengthen industry standards, norms, and self-disciplinary mechanisms that direct and guide service providers to develop standards, comply with requirements, and accept social oversight.

第六条 Content and harm prevention

Providers must maintain mainstream values and prevent harmful content. They are prohibited from using recommendation services for activities that endanger national security, public interest, or legal rights, and must prevent dissemination of prohibited information and harmful content.

第七条 Management systems and controls

Providers must establish responsibility for algorithm safety and implement systems and measures including algorithm review, ethics assessment, user registration, content audit, data and privacy protection, fraud prevention, security monitoring, and incident response. They must disclose service rules and maintain adequate personnel.

第八条 Algorithm evaluation and verification

Providers must regularly audit, evaluate, and verify their algorithms, models, data, and results. They are prohibited from deploying algorithms that induce user addiction, excessive consumption, or behavior that violates laws or ethical standards.

第九条 Information security and harmful content handling

Algorithm recommendation service providers must strengthen information security, establish a system to detect illegal and harmful content, and apply clear labeling to machine-generated synthetic information before distributing it. Providers must immediately stop transmitting illegal information and report it to regulatory authorities.

第十条 User model and tag management

Algorithm recommendation service providers must manage user profiles and tags properly, ensuring they do not record keywords related to illegal or harmful content as user interests or tags that would be used for content distribution.

第十一条 Feed page ecology management

Algorithm recommendation service providers must manage the layout and presentation of their platforms, establish human oversight and user choice mechanisms, and actively display mainstream-value-aligned information at key locations such as homepages and popular trends.

第十二条 Transparency and user protection measures

Algorithm recommendation service providers are encouraged to apply content deduplication and distribution strategies while improving the clarity and explainability of their filtering and ranking rules, to prevent harm and reduce disputes.

第十三条 Internet news information services

Providers offering news services through algorithm recommendations must obtain proper licenses, follow legal standards for news collection and distribution, and must not generate false news or distribute news from unauthorized sources.

第十四条 Prohibited algorithmic manipulation practices

Algorithm recommendation service providers are prohibited from using algorithms to falsely register accounts, manipulate user engagement, block information, manipulate rankings or search results, control trending topics, or interfere with information presentation in ways that affect public opinion or evade oversight.

第十五条 Prohibition on anticompetitive behavior

Algorithm recommendation service providers must not use algorithms to unreasonably restrict other internet service providers or disrupt their normal operations, and must not engage in monopolistic or unfair competition practices.

第十六条 Transparency disclosure requirements

Algorithm recommendation service providers must clearly inform users that they use algorithmic recommendations and must publicly disclose the basic principles, purposes, and main operational mechanisms of their algorithms.

第十七条 User choice and control options

Algorithm recommendation service providers must offer users a non-personalized option or an easy way to disable algorithmic recommendations, and must immediately stop providing the service when users choose to disable it. Providers must also allow users to view and delete their personal profile tags, and must explain and be responsible when algorithms significantly affect user rights.

第十八条 Services for minors

Providers of algorithmic recommendation services must fulfill legal obligations to protect minors online, develop age-appropriate modes, and curate information beneficial to their physical and mental wellbeing. They must not push content that could prompt unsafe behavior, violate social norms, induce harmful habits, or lead to internet addiction among minors.

第十九条 Services for elderly users

Providers of algorithmic recommendation services must safeguard the legitimate rights of older adults by considering their needs in transportation, healthcare, consumption and administrative matters, offering elderly-adapted intelligent services according to national standards, and monitoring and handling telecommunications fraud information to enable safe use.

第二十条 Work scheduling services

Providers offering work dispatch through algorithmic recommendations must protect workers' entitlements to compensation, rest and leave by creating sound algorithms governing order allocation, payment composition and timing, working hours, and reward or penalty mechanisms.

第二十一条 Consumer protection in sales

Providers offering goods or services to consumers through algorithmic recommendations must protect the right to fair dealing and must not use algorithms to apply discriminatory pricing or other unreasonable transaction conditions based on consumer preferences or purchase history.

第二十二条 User complaints and grievances

Providers of algorithmic recommendation services must establish simple and effective channels for user complaints and public complaints or reports, specify processing procedures and response timeframes, and promptly handle and report results back to users.

第二十三条 Tiered classification system

Relevant government departments shall establish a graded and categorized safety management framework for algorithmic recommendations based on the service's potential to influence public opinion or mobilize society, content types, user scale, data sensitivity, and degree of intervention in user behavior.

第二十四条 Registration requirements

Providers with algorithmic services having public influence or social mobilization capacity must register within ten working days, submitting information including provider name, service type, application domain, algorithm type, self-assessment reports and proposed public disclosures through an official system. They must update registrations within ten working days of changes and file for deregistration within twenty working days of service termination.

第二十五条 Registration approval

Relevant government departments shall review complete registration materials and issue approval with a registration number within thirty working days, or inform the applicant of rejection with reasons within that timeframe.

第二十六条 Public disclosure of registration

Providers that complete registration must clearly display their registration number on their websites, applications and other service platforms and provide links to the public information about their registration.

第二十七条 Safety assessment requirement

Providers of algorithmic recommendation services with news attributes or social mobilization capacity must conduct safety evaluations according to national regulations.

第二十八条 Regulatory oversight and compliance

Cyberspace authorities and related departments conduct safety evaluations and inspections of algorithmic services, issue correction notices with deadlines, and require providers to retain network logs and cooperate by providing technical and data support.

第二十九条 Confidentiality of personal and commercial information

Personnel involved in safety assessments and inspections must keep confidential any personal privacy, personal information, and business secrets they learn about during their work and must not disclose them.

第三十条 Public complaint and reporting mechanism

Any organization or individual may report violations of this regulation to cyberspace authorities and relevant departments, which must promptly handle such reports according to law.

第三十一条 Penalties for major violations

Providers violating specified sections face warnings, public criticism, orders to correct within deadlines, and fines of 10,000 to 100,000 yuan if they refuse to comply or the violation is serious; applicable criminal or administrative penalties also apply.

第三十二条 Penalties for other violations

Providers violating other specified sections face administrative penalties determined by relevant authorities according to applicable laws, regulations, and departmental rules.

第三十三条 Penalties for fraudulent registration and termination violations

Providers with news or social mobilization attributes that fraudulently obtain registration face deregistration, warnings, and fines up to 100,000 yuan; providers that fail to properly deregister or face closure orders may have their registration revoked by cyberspace authorities.

第三十四条 Interpretation authority

The Cyberspace Administration together with industrial, information technology, public security, and market supervision departments is responsible for explaining this regulation.

第三十五条 Effective date

The regulation enters into force on March 1, 2022.