Appendix B. Extended Security Control Tables

This appendix provides expanded reference tables (Tables B.1–B.10) that catalog additional security controls applicable across the phases of the AI lifecycle. These tables are intended to complement the prioritized controls presented in the main body of this guide by offering a more comprehensive view of relevant safeguards.

For each AI lifecycle phase, the tables list individual security controls, identify the Open Web Application Security Project (OWASP) and Berryville Institute of Machine Learning (BIML) threat categories that each control helps mitigate, and explain the rationale for why each control is relevant to that phase. Together, this information enables readers to understand not only what controls should be applied, but also which risks they address and why these controls are important.

These controls are not exhaustive or mandatory; they serve as supplemental reference to support risk-based decisionmaking. Organizations may select, adapt, or prioritize controls based on their specific AI use cases, systems, threat models, operational constraints, and regulatory obligations.

• Design
  • Table B.1. Security Controls Related to Data
  • Table B.2. Security Controls Related to Human Talent
  • Table B.3. Security Controls Related to Compute (on Premises)
  • Table B.4. Security Controls Related to Compute (Cloud)
  • Table B.5. Security Controls Related to Algorithms
• Develop
  • Table B.6. Security Controls Related to Model Training
  • Table B.7. Security Controls Related to Evaluation
• Deploy
  • Table B.8. Security Controls Related to Model Weights
• Operate
  • Table B.9. Security Controls Related to Integration into Internal Products
  • Table B.10. Security Controls Related to Integration and External Products