AI-Assisted Secure Coding (Appendix C) · AI Tool Qualification & Threat Modeling
Use a recognised evaluation framework
Verify that evaluations follow a recognized framework such as NIST AI RMF, NIST AI 600-1 Generative AI Profile, or ISO/IEC 42001. Evaluations are repeated after a major version change, a vendor incident, or new threat intelligence relevant to the tool class.
OWASP AISVS v1.01, requirement AC.2.4
In plain English
Base your tool evaluations on a recognised framework, such as the NIST AI Risk Management Framework, its Generative AI Profile (NIST AI 600-1) or ISO/IEC 42001. Repeat them after a major version change, a vendor incident or relevant new threat intelligence. This is a Level 3 control.
What good looks like
- Evaluations follow a named framework and cover its relevant parts.
- Re-evaluation triggers are defined: a major version, a vendor security incident, or new threat intelligence for that class of tool.
- Records show when evaluations were repeated and why.
- Someone is responsible for watching for the triggers.
How teams usually do it
- Map your evaluation template to the chosen framework's functions or clauses, so gaps are visible.
- Subscribe to vendor security advisories and changelogs, and record the trigger review in the vendor management process.
Suggested evidence
- Evaluation records with dates and triggers.
- The mapping from your template to the framework.
Common mistakes
- Citing a framework without following it.
- Defining triggers and having nobody watch for them.
Not sure which requirements apply to your system? Take the assessment, or browse every guide in this chapter.
Requirement text from the OWASP AI Security Verification Standard, licensed under CC BY-SA 4.0. This guidance is shared under the same licence.