China: Technical Documentation of National Technical Committee 260 on Cybersecurity of Standardization Administration of China: Basic Safety Requirements for Generative Artificial Intelligence Services

1. 5.1 Corpus Source Safety Requirements 3
2. 5.2 Corpus Content Safety Requirements 4
3. 5.3 Corpus Annotation Safety Requirements 5
4. 8.1 Keyword Library 10
5. 8.2 Generated Content Test Question Bank 10
6. 8.3 Refusal to Answer Test Question Bank 11
7. 8.4 Classification Models 12
8. 9.1 Assessment Methods 12
9. 9.2 Corpus Safety Assessment 14
10. 9.3 Generated Content Safety Assessment 14
11. 9.4 Assessment of Refusal to Answer Questions 14
12. 1 Scope
13. 2 Normative Reference Documents
14. 3 Terminology and Definitions
15. 3.1 Generative Artificial Intelligence Services
16. 3.2 Service Provider
17. 3.3 Training Data (训练语料)
18. 3.4 Sampling Qualified Rate (抽样合格率)
19. Appendix A of this document.
20. 3.5 Foundation Model
21. 3.6 Illegal and unhealthy information (违法不良信息)
22. 4 General Provisions
23. 5 Corpus Safety Requirements
24. 5.1 Corpus Source Safety Requirements
25. 5.2 Corpus Content Safety Requirements
26. 5.3 Corpus Annotation Safety Requirements
27. 6 Model Safety Requirements
28. 7 Safety Measure Requirements
29. 8 Other Requirements
30. 8.1 Keyword Library
31. Appendix A.2.: c) The keyword library shall be updated in a timely manner in accordance
32. 8.2 Generated Content Test Question Bank
33. Appendix A.1 and A.2, and there should be no fewer than 20 test
34. 8.3 Refusal to Answer Test Question Bank
35. 8.4 Classification Models
36. 9 Safety Assessment Requirements
37. 9.1 Assessment Methods
38. 9.2 Corpus Safety Assessment
39. 9.3 Generated Content Safety Assessment
40. 9.4 Assessment of Refusal to Answer Questions
41. Appendix A Main Safety Risks of Corpora and Generated Content

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