New-Generation AI Governance Principles
📑 Legal hierarchy: Level 4 · Normative document (governance principles) | Issuance: National New-Generation AI Governance Expert Committee (convened by MOST) | Released: 2019-06-17 | Character: soft law · value anchor
⚠️ Hierarchy note: This document is a statement of governance principles (Level 4 normative document); it does not directly impose legal duties, but provides directional anchoring for every subsequent AI departmental rule. See Index of Chinese Rules.
Chinese Summary
Section titled “Chinese Summary”The New-Generation AI Governance Principles — Developing Responsible AI were issued by the National New-Generation AI Governance Expert Committee on 2019-06-17. This is China’s first dedicated official AI governance-principles document.
Significance: before the subsequent AI departmental rules (the Generative AI Interim Measures, Deep Synthesis Provisions, etc.), these Principles provided the basic value anchor for Chinese AI governance; phrasings in later rules (e.g., “balance development and security”) can be traced to them.
The Eight Principles
Section titled “The Eight Principles”- Harmony and Friendship (和谐友好)
- Fairness and Justice (公平公正)
- Inclusion and Sharing (包容共享)
- Respect for Privacy (尊重隐私)
- Safety and Controllability (安全可控)
- Shared Responsibility (共担责任)
- Open Cooperation (开放协作)
- Agile Governance (敏捷治理) — first official articulation of “agile governance”.
Follow-on Developments
Section titled “Follow-on Developments”Ethics Norms for a New Generation of AI (2021-09)
Section titled “Ethics Norms for a New Generation of AI (2021-09)”The same expert committee released the more granular Ethics Norms for a New Generation of AI on 2021-09-25, setting out 6 basic ethical requirements and 18 specific ethical requirements.
Impact on Departmental Rules
Section titled “Impact on Departmental Rules”- “Agile governance” → the institutional backdrop for instruments with the “Interim” (暂行) naming.
- “Safety and controllability” → the algorithm filing / security-assessment regime.
- “Respect for privacy” → alignment with PIPL.
- “Shared responsibility” → the tripartite liability structure of service provider + user + platform.
Comparison with International Governance Principles
Section titled “Comparison with International Governance Principles”- OECD AI Principles (2019-05), contemporaneous: transparency, fairness, robustness, accountability, human-centric, inclusion.
- G20 AI Principles (2019-06 Osaka Declaration), aligned with OECD.
- Chinese Principles vs. OECD: substantial overlap on value axes, but “shared responsibility” and “inclusion and sharing” emphasize the collective dimension more strongly; OECD places more emphasis on human rights and autonomy.
As Diplomatic Expression for Global Governance
Section titled “As Diplomatic Expression for Global Governance”In 2023-10, China went further with the Global AI Governance Initiative, an externally facing complement to these Principles and the position document China advances in the UN, bilateral, and multilateral contexts.
Source Text and Archival Copies
Section titled “Source Text and Archival Copies”| Source | Link |
|---|---|
| Chinese (MOST) | most.gov.cn |
| English translation | China Law Translate |
Version History
Section titled “Version History”| Date | Event |
|---|---|
| 2019-06-17 | Eight governance principles released |
| 2021-09-25 | Ethics Norms for a New Generation of AI (6 + 18) released |
| 2023-10 | Global AI Governance Initiative (external-facing) |
| 2024-09 | AI Safety Governance Framework 1.0 (from principles to a systematic framework) |