Add architectural note to C9.6: evaluation-only policy components#689
Closed
emmanuelgjr wants to merge 1 commit intoOWASP:mainfrom
Closed
Add architectural note to C9.6: evaluation-only policy components#689emmanuelgjr wants to merge 1 commit intoOWASP:mainfrom
emmanuelgjr wants to merge 1 commit intoOWASP:mainfrom
Conversation
Adds an architectural note in C9.6 clarifying that policy decision components for agent authorization should be evaluation-only and not possess execution capabilities, following the NIST SP 800-207 control-plane/data-plane separation principle. This is an architectural note rather than a standalone control per reviewer consensus, as the constraint is already implied by C9.6.4, C5.6.4, and C5.2.6 combined. Closes OWASP#675
Collaborator
|
After thinking about it... @ottosulin / @jmanico do we need an architectural note for a section? Or would that be too much for the people using the AISVS? |
Collaborator
|
I think we should keep an eye out for a scope note rather than an architectural note, even if it was my first intent to do that. For now, I would close this PR and this issue, and we'll see if this reqs are still there after the reduction pass of C9. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Rationale
The original proposal (#675) suggested a standalone verifiable control requiring governance components to not possess execution ability. Reviewer feedback identified two issues: (1) the constraint is not clearly verifiable as a standalone control (an auditor cannot reliably enumerate all execution pathways), and (2) it is not AI-specific — control-plane/data-plane separation is a general NIST SP 800-207 principle.
The existing C9.6.4, C5.6.4, and C5.2.6 already push implementers toward purpose-built evaluation components. An architectural note makes this design expectation explicit without creating a control that is hard to audit independently.
Test plan
Closes #675