How to meet it
OWASP AISVS v1.01
Adversarial Robustness · Model Alignment, Safety, and Robustness Testing and Training
Report adversarial results alongside normal ones
Verify that adversarial evaluation results are reported alongside results for the same tasks, scored the same way, with no attack applied.
OWASP AISVS v1.01, requirement 11.1.6
In plain English
Report results under attack next to results for the same tasks without attack, scored the same way, so the size of the drop is clear.
What good looks like
- Each adversarial result has a matching clean result.
- Scoring is identical.
- Reports show both together.
How teams usually do it
- Running each evaluation task in clean and attacked form in the same job.
Evidence an auditor would accept
- Evaluation reports showing both columns.
Common mistakes
- Reporting attack success rates with no baseline, so nobody can tell what they mean.
Not sure which requirements apply to your system? Take the assessment.
Requirement text from the OWASP AI Security Verification Standard, licensed under CC BY-SA 4.0. This guidance is shared under the same licence.