China: National Standard of the People's Republic of China: Cybersecurity Technology—Safety Specifications for Generative Artificial Intelligence Pre-Training and Fine-Tuning Data (Draft for Feedback)

1. Appendix A (for reference) Main Safety and Security Risks of Pre-Training and Fine-
2. 1. Scope
3. 2. Normative References
4. 3. Terminology and Definitions
5. 3.1 Generative Artificial Intelligence
6. 3.2 GenAI Services
7. 3.3 Service Provider
8. 3.4 Service User
9. 3.5 Pre-Training
10. 3.6 Fine-Tuning
11. 3.7 Pre-Training Data
12. 3.8 Fine-Tuning Data
13. 4. Overview
14. 4.1 Safety and Security Risks
15. 4.2 Safety and Security Framework
16. 5. General Safety and Security Requirements
17. 6. Safety and Security Requirements for Pre-Training Data Processing Activities
18. 6.1 Data Collection
19. 6.2 Data Pre-Processing
20. 6.3 Data Usage
21. 7. Safety and Security Requirements for Fine-Tuning Data Processing Activities
22. 7.1 Data Collection
23. 7.2 Data Pre-Processing
24. 7.3 Data Usage
25. 8. Evaluation Methods
26. 8.1 General Safety and Security Evaluation Methods
27. 8.2 Evaluation Methods for Pre-Training Data Processing Activities
28. 8.2.1 Data Collection
29. 8.2.2 Data Pre-Processing
30. 8.2.3 Data Usage
31. 8.3 Evaluation Methods for Fine-Tuning Data Processing Activities
32. 8.3.1 Data Collection
33. 8.3.2 Data Pre-Processing
34. 8.3.3 Data Usage
35. Appendix A
36. Appendix B
37. Appendix A.1, and there shall be no fewer than 100 keywords for each safety

The outline comes from headings detected in the text, so it might miss some or include extras.