How to meet it
OWASP AISVS v1.01
Model Behavior, Output Control & Safety Assurance · Output Format Enforcement
Don't act when the AI refuses or fails
Verify that, when an application uses an AI response to decide whether to perform an action, it does not perform that action if the AI refuses the request, reports a failure, or returns a response the application cannot correctly read or understand.
OWASP AISVS v1.01, requirement 7.1.3
In plain English
If your software uses the model's answer to decide whether to act, it must not act when the model refuses, reports an error or returns something the software cannot parse.
What good looks like
- Actions run only on a clear, valid, positive result.
- Refusals, errors and unparseable responses lead to no action.
- These cases are logged.
How teams usually do it
- Explicit success fields in the output schema.
- Code paths where the default is to do nothing.
Evidence an auditor would accept
- Code showing the decision logic.
- Tests with refusals and garbled output.
Common mistakes
- Treating any non-empty response as approval.
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.