China Ministry of Industry and IT: Measures for AI Technology Ethics Management Services (Draft)

Summary

Regulates artificial intelligence (AI) scientific and technological activities within China to address ethical risks and challenges, emphasizing fair, just, safe, and responsible innovation.

Establishes AI technology ethics management services, requiring the integration of ethical principles like promoting human well-being, respecting life and intellectual property rights, ensuring fairness, transparency, and accountability.

Promotes a system of AI technology ethics standards and international cooperation in their formulation.

Encourages institutions to engage in innovation in AI ethics and to conduct public awareness campaigns, education, and training.

Requires the establishment of ethics committees and service centers for AI ethics reviews, with procedures for general, simplified, or emergency reviews depending on risk levels.

Designates government ministries to oversee AI ethics governance and ensures violations are subject to legal penalties.

Encourages collaboration and mutual recognition of ethics review results among entities.

Regulates AI ethical governance to promote responsible innovation and healthy industry development in China.

(Trial) Measures for Artificial Intelligence Technology Ethics Management Services (Draft for Public Feedback)

Chapter 1 General Provisions

Article 1 [Purpose and Basis] These Measures have been formulated in accordance with the Law of the People’s Republic of China on Progress of Science and Technology, the Opinions on Strengthening the Governance of Science and Technology Ethics, and the (Trial) Measures for Scientific and Technological Ethics Reviews (hereinafter referred to as the Ethics Measures), and other laws, regulations, and relevant provisions, in order to regulate the ethical governance of artificial intelligence (AI) scientific and technological (S&T) activities, promote fair, just, harmonious, safe, and responsible innovation, and spur the healthy development of the AI industry.

Requires AI S&T activities in China to conduct ethics review, adhering to specified ethical principles.

Article 2 [Scope of Application] The AI S&T activities referred to in these Measures are those AI-related S&T activities conducted within the People’s Republic of China that may pose ethical risks and challenges in areas such as human health, human dignity, the ecological environment, public order, and sustainable development, including AI scientific research and technology development, as well as other S&T activities subject to AI S&T ethics review in accordance with laws, administrative regulations, and relevant national provisions.

Article 3 [Ethical Principles] Those carrying out AI S&T activities shall incorporate S&T ethics requirements throughout the entire process, and shall adhere to the AI S&T ethics principles of promoting human well-being, respecting the right to life, respecting intellectual property rights, upholding fairness and justice, reasonably controlling risks, maintaining openness and transparency, ensuring controllability and trustworthiness, and strengthening accountability, and shall comply with China’s Constitution, laws, regulations, and relevant provisions.

Establishes and promotes AI ethics standards and service systems for risk monitoring, assessment, and international cooperation.

Chapter 2 Support and Promotion of AI Technology Ethics Article 4 [Standards Construction] A system of AI technology ethics standards shall be established and improved; the formulation of international, national, industry, and group standards for AI technology ethics shall be promoted; and the establishment of platforms for international standardization exchange and cooperation shall be supported. Institutes of higher education, research institutions, medical and health institutions, enterprises, and S&T-related social organizations are encouraged to participate in the formulation, validation, and promotion of AI technology ethics standards.

[footnotes omitted]

Article 5 [Service System Mechanisms] Development of an AI technology ethics service system shall be promoted; service provision shall be strengthened in areas such as risk monitoring and early warning, detection and assessment, certification, and consultation; the capabilities of enterprises in technology R&D and prevention of AI technology ethical risks shall be improved; support for micro-, small, and medium-sized enterprises in the area of AI technology ethics shall be expanded; and nternational exchange and cooperation in AI technology ethics shall be promoted.

Encourages innovation and education in AI ethics, promoting open-sourcing, risk assessment tools, and public awareness.

Article 6 [Encouraging Innovation] Institutes of higher education, research institutions, medical and health institutions, enterprises, and S&T-related social organizations shall be encouraged to engage in cutting-edge technological innovation and cross-disciplinary research in the field of AI technology ethics. Technological innovation in AI technology ethics shall be supported, and the use of technical means to prevent AI technology ethical risks shall be strengthened. The orderly open-sourcing and sharing of AI S&T ethics technologies, intellectual property, and high-quality datasets shall be promoted. Research and development of general-purpose risk management, assessment, and auditing tools shall be enhanced, and scenario-based S&T ethical risk assessment and evaluation shall be explored. AI products and services that comply with S&T ethics shall be promoted, and the intellectual property rights of S&T ethics-related technologies shall be protected.

Article 7 [Propaganda and Education] Relevant work units (单位) shall be guided to conduct public awareness campaigns and education on AI technology ethics. The role of S&T ethics-related social organizations in such efforts shall be fully leveraged, public participation shall be encouraged, and practical demonstrations shall be promoted. Mass media shall be encouraged to conduct targeted propaganda and education on AI technology ethics.

[footnotes omitted]

Article 8 [Talent Cultivation] Institutes of higher education, research institutes, medical and health institutions, enterprises, and S&T-related social organizations shall be supported in conducting education and training on AI technology ethics. The development of vocational systems and curricula shall be promoted, various means shall be used to develop AI technology ethics talent, and talent exchange shall be promoted.

Designates work units responsible for AI ethics management; encourages establishing ethics committees and service centers.

Chapter 3 Implementing Entities

Article 4 [Entities] Work units that engage in AI S&T activities, including institutes of higher education, scientific research institutions, medical and health institutions, and enterprises, shall be the entities responsible for AI technology ethics management services within their respective work units. Work units with the necessary prerequisites shall establish Artificial Intelligence Technology Ethics Committees (hereinafter referred to as “Committees”). Committees shall be provided with the necessary staff, office space, funding, and other requirements, and effective measures shall be taken to assure that Committees carry out their work independently. Qualified relevant work units are encouraged to conduct certification of AI technology ethics management systems.

Article 10 [Duties of Committees and Committee Members] A Committee’s charter and composition, and the duties and obligations of its members, shall be in accordance with Articles 5 through 8 of the Ethics Measures. The composition of Committee shall include experts with relevant professional backgrounds in AI technology, applications, ethics, and law.

Article 11 [Ethics Service Centers] Local and relevant main oversight departments (主管部门) may establish specialized AI technology ethics service centers (hereinafter referred to as “Service Centers”), taking into account actual conditions and relying on relevant work units. Service Centers will accept commissions from other work units to provide them services such as ethics reviews of AI S&T activities, training, and consultation. Service Centers shall establish regulated management systems and procedures, employ full-time staff with the ability to provide AI technology ethics services, and accept supervision by the local or relevant main industrial oversight department.

Requires AI S&T activity leaders to apply for ethics review and submit comprehensive application materials.

Chapter 4 Work Procedures Section 1 Application and Acceptance Article 12 [Application] For AI S&T activities falling within the scope indicated in Article 2 of these Measures, the person in charge of the AI S&T activity shall apply to the Committee of their work unit. If the organization has not established a Committee or if the Committee is unable to meet the requirements for AI technology ethics reviews, the person in charge shall apply to the Service Center commissioned by the work unit to conduct the AI technology ethics review. If the person in charge is not affiliated with any work unit, they shall commission a qualified Service Center to conduct the S&T ethics review. The person in charge of the AI S&T activity shall submit application materials to the Committee or Service Center in accordance with regulations. The application materials shall mainly include: (1) The AI S&T activity proposal, including the research background, objectives, plan, and implementation steps; documentation of the lawful capabilities (合法资质) of the relevant institutions involved; personnel details; funding sources; the mechanisms and principles of the algorithms to be used; the types, sources, and methods of data acquisition and cleaning; testing and evaluation methods; the hardware and software products to be developed; and the anticipated fields of application and user groups;

(2) An S&T ethical risk assessment and risk prevention and control plan for the AI S&T activity, including an assessment of potential S&T ethical risks arising from the intended use, misuse, and abuse of AI technology, as well as plans for the prevention and control of S&T ethical risks stemming from the activity; (3) A written commitment to comply with AI technology ethics and research integrity requirements.

Requires the Committee or Service Center to determine AI ethics review procedures based on S&T ethical risks.

Article 13 [Acceptance Notification and Selection of Procedures] The Committee or Service Center shall decide whether to accept the application based on the submitted materials, and notify the applicant. If the application is accepted, the Committee or Service Center shall specify the applicable procedures—general, simplified, or emergency—based on the likelihood and severity of S&T ethical risks and the urgency of the situation. Based on the requirements of the selected procedures, various formats shall be employed to conduct the S&T ethics review, including in-person meetings, online meetings, or written inquiry. If the materials are incomplete, notification of all required supplementary materials shall be made in a single communication.

Section 2 General Procedures and Simplified Procedures Article 14 [Implementation of General Procedures] The AI technology ethics review meeting shall be presided over by the Committee chair or his or her designated deputy chair. There shall be no fewer than 5 members present, and they shall include members from the different categories listed in Article 10 of these Measures. The Service Center may form an expert group to organize and implement the review in accordance with the Committee’s regulations. Based on the requirements of the review, advisory experts from relevant fields who do not have a direct interest may be invited to provide advisory opinions. Advisory experts shall not participate in voting at the meeting.

Requires AI ethics reviews to focus on fairness, controllability, transparency, accountability, and qualified personnel capabilities.

Article 15 [Review Focus] When a Committee or Service Center conducts an AI technology ethics review, the focus shall be on the following aspects: (1) Fairness and impartiality: Adopt effective measures to enhance data completeness and diversity; ensure the rational design of algorithms, models, and systems; prevent bias, discrimination, and algorithmic exploitation; and assure the objectivity and inclusivity of resource allocation, access to opportunities, and decision-making processes. (2) Controllability and trustworthiness: Adopt robust design and other relevant measures to assure that models and systems can withstand open environments, extreme conditions, and disruptive factors, and that end-users can control, guide, and intervene in the basic operations of models and systems.Continuous monitoring and operation: Timely remedial measures can be taken in the event of emergencies. (3) Transparency and explainability: Make timely disclosures of relevant information, including but not limited to the purpose, operational logic, interaction methods, and potential risks of algorithms, models, and systems. Employ effective technical means to increase the explainability of algorithms, models, and systems. (4) Accountability and traceability: Implement measures such as log management to clearly record relevant information at every stage of data, algorithms, models, and systems, and assure end-to-end traceability and management, so that specific stages can be precisely identified and responsible parties determined when issues arise. (5) The capabilities of scientific and technical personnel who are to conduct AI S&T activities must meet relevant requirements. Such activities must possess scientific and social value, and the research objectives must play a positive role in advancing human well-being and achieving the sustainable development of society.The AI S&T activity shall have reasonable risks relative to benefits, and the S&T ethical risk control plan and emergency plan shall be scientific, appropriate to the circumstances, and operationally feasible.

Requires the Committee or Service Center to review applications, provide decisions, and handle appeals for AI activities.

Article 16 [Application Handling and Decisions] The Committee or Service Center shall, within 30 days of accepting the application, make a decision to approve, review again after revision, or deny the application. In special circumstances such as where the matter is complex or where supplementary or corrected materials are required, the time limit may be extended appropriately, with the extended time limit clearly specified. Where an application requires revision or is denied, the Committee or the Service Center shall provide recommendations for revision or explain its reasons. If the applicant objects to the decision, they shall file an appeal with the Committee or Service Center within 3 working days from the date of receipt of the decision. If the grounds for appeal are well-founded, the Committee or Service Center shall make a new decision within 7 working days.

Article 17 [Follow-up Management] Persons in charge of AI S&T activities shall promptly identify changes in S&T ethical risks and report such changes to the Committee or Service Center. The Committee or Service Center shall conduct follow-up reviews of approved AI S&T activities in accordance with the provisions of Article 19 of the Ethics Measures, so as to keep abreast of changes in S&T ethical risk circumstances, and where necessary it may decide to suspend or terminate the relevant S&T activities. The interval before a follow-up review shall generally not exceed 12 months.

Article 18 [Mutual Recognition of Results] Where multiple work units collaborate on an AI S&T activity, mutual recognition among work units of AI S&T ethics review results may be implemented, depending on the actual circumstances.

Allows simplified procedures for AI activities with ethical risks similar to daily life or minor modifications.

Article 19 [Applicable Circumstances for Simplified Procedures] Simplified procedures may be applied in any of the following circumstances: (1) Where the likelihood and severity of the AI S&T activity’s ethical risks do not exceed the conventional risks encountered in daily life; (2) Where minor modifications are made to an approved AI S&T activity plan and they do not increase the risk-benefit ratio; (3) Where a follow-up review is for an AI S&T activity that has not undergone major adjustments in the preceding period.

Article 20 [Implementation of Simplified Procedures] The Committee or Service Center shall establish working procedures and follow-up frequencies applicable to reviews under simplified procedures. A review under simplified procedures shall be undertaken by at least two members designated by the Committee chair. The Service Center may organize and implement such reviews in accordance with the Committee’s stipulations. Where, during a review under simplified procedures, the review result is a negative opinion, there are concerns regarding the review content, or there is a disagreement among committee members, general procedures shall be applied instead.

Requires MIIT, MOST to list AI activities needing ethical reconsideration and mandates expert review applications.

Section 3 Expert Reconsideration Procedures Article 21 [Reconsideration List] The Ministry of Industry and Information Technology (MIIT) and the Ministry of Science and Technology (MOST), in conjunction with relevant departments, shall formulate and publish a “List of AI S&T activities for which expert reconsideration of S&T ethics must be conducted” (hereinafter referred to as the “Reconsideration List”), which shall be dynamically adjusted as needed.

Article 22 [Application for Reconsideration] For AI S&T activities included in the “Reconsideration List,” after the preliminary review by the Committee or Service Center has been passed, the work unit concerned shall apply for expert reconsideration.Where multiple work units are involved, the lead work unit shall be responsible for making the application. Central enterprises,6 as well as institutes of higher education, research institutions, and medical and health institutions directly under Central Committee and State agencies, shall submit applications directly to the relevant main oversight department to organize and conduct expert reconsideration. Other work units shall submit applications to the local authorities to lead the carrying out of expert reconsideration.

Requires oversight departments to establish a panel for AI ethics review reconsideration, providing feedback in 30 days.

Article 23 [Implementation of Reconsideration] The work unit undertaking the AI S&T activity shall submit materials for expert reconsideration in accordance with the provisions of Article 27 of the Ethics Measures. Local or relevant main oversight departments shall organize and establish an expert reconsideration panel in accordance with the provisions of Articles 28 through 30 of the Ethics Measures to reconsider the compliance and reasonableness of the preliminary review opinion, and shall provide feedback on the reconsideration opinion to the applying work unit within 30 days of receiving the application for reconsideration. [footnotes omitted]

The local or relevant main oversight department may entrust the Service Center with the specific implementation of the reconsideration work.

Article 24 [Application of Results] The Committee or Service Center shall make its S&T ethics review decision based on the expert consideration opinion.

Article 25 [Post-Reconsideration Follow-Up] The Committee or Service Center shall strengthen follow-up reviews of AI S&T activities included in the“Reconsideration List,” with the intervals between follow-up reviews generally not to exceed 6 months. Where major changes in S&T ethical risks have occurred, relevant S&T ethics reviews shall be conducted again in accordance with the provisions of Article 20 of the Ethics Measures, and an application shall be made for expert reconsideration.

Article 26 [Coordination with Other Regulatory Measures] Where regulatory measures such as registration, filing, or administrative approval are implemented for AI S&T activities in areas including deep synthesis, algorithmic recommendations, and generative AI service management, and compliance with S&T ethics requirements is a condition for approval and a component of regulatory oversight, expert reconsideration may be dispensed with.

Requires the Committee or Service Center to establish and oversee an emergency review system for AI incidents.

Section 5 Emergency Procedures Article 27 [Emergency Work System] The Committee or Service Center shall formulate a system for emergency review work, specifying the emergency review process and standard working procedures in emergencies such as sudden public security incidents (突发公共事件). Emergency review work shall generally be completed within 72 hours. For AI S&T activities subject to expert reconsideration procedures, the preparatory work prior to expert reconsideration shall generally be completed within 36 hours.

Article 28 [Conduct of Emergency Procedures] The Committee or Service Center shall ensure the quality and timeliness of S&T ethics-related work, and strengthen follow-up work and process supervision. When necessary, advisory experts from relevant fields may be invited to attend meetings to provide advisory opinions.

Assigns MOST to lead S&T ethics supervision; MIIT governs AI technology ethics; departments manage within jurisdictions.

Chapter 5 Supervision and Management Article 29 [Division of Responsibilities] MOST is responsible for overall guidance of S&T ethics supervision nationwide. MIIT, in conjunction with relevant departments, is responsible for the governance of AI technology ethics and for strengthening the coordination and guidance of emergency ethics work. Each such department shall be responsible for AI technology ethics management and service work within its own industry or system, in accordance with its respective duties and authority. Each local agency shall be responsible for AI technology ethics management services work within its own jurisdiction, in accordance with its respective duties and authority.

Article 30 [Registration and Submission of Materials] Work units shall, in accordance with the provisions of Articles 43 to 45 of the Ethics Measures, register relevant Committee information and AI S&T activities included in the “Reconsideration List” through the National Science and Technology Information System Public Service Platform, and submit Committee annual work reports, implementation reports on AI S&T activities included in the “Reconsideration List,” and other relevant materials.Service centers shall register and submit their annual work reports in accordance with the aforementioned provisions. MOST and relevant main oversight departments shall share registration information related to AI technology ethics in a synchronized manner.

Requires oversight departments to maintain reporting channels for AI ethics violations and handle them per regulations.

Article 31 [Reporting of Issues] Local and relevant main oversight departments and work units engaged in AI S&T activities, shall, taking into account the actual circumstances of their respective industries, systems, or work units, maintain effective channels for reporting issues involving violations of AI technology ethics laws and regulations, and handle such matters in accordance with relevant provisions.

Article 32 [Handling of Violations] Violations of the provisions of these Measures during the course of AI S&T activities or the conduct of work related to AI technology ethics shall be subject to corresponding punishments in accordance with the Law on Progress of Science and Technology and other laws, regulations, and relevant provisions.

Defines "local" and "relevant main oversight department" for AI S&T ethics management and oversight.

Chapter 6 Supplementary Provisions

Article 33 [Definition of terms] Where provisions in these Measures involve a time period and working days are not indicated, it refers to natural days.As used in these Measures, the term “local” refers to the provincial-level administrative department responsible for the review and management of S&T ethics in the field of AI as determined by the local people’s government at the provincial level, and the term “relevant main oversight department” refers to the relevant main oversight department of the State Council.

Article 34 [Complementary Rules] Local and relevant main oversight departments shall formulate and amend rules and regulations in their respective jurisdictions, industries, and systems, such as measures and detailed rules for AI technology ethics management services, in accordance with the provisions of these Measures and taking into account their actual circumstances. S&T social organizations may formulate specific rules and guidelines for S&T ethics reviews in their respective fields.

Implements special AI ethics provisions if consistent with broader measures; assigns MIIT for interpretation.

Article 35 [Special Provisions] Where relevant main industrial oversight departments have special provisions regarding AI technology ethics management services within their industries or systems, and such provisions are consistent with the spirit of these Measures, implementation shall be in accordance with those provisions as well. For matters not stipulated in these Measures, implementation shall be in accordance with the Ethics Measures and the provisions of relevant laws and regulations.

Article 36 [Interpretation of these Measures] MIIT, in conjunction with relevant departments, shall be responsible for the interpretation of these Measures.

Article 37 [Date of implementation] These Measures shall come into force on ____________________(date).

[appendix omitted]