How to meet it
OWASP AISVS v1.01
Input Validation · Content & Policy Screening
Check images, audio and video for hidden attacks
Verify that non-text inputs (image/video/audio) are checked for adversarial perturbations, steganographic payloads, hidden or embedded content, or known attack patterns.
OWASP AISVS v1.01, requirement 2.2.3
In plain English
If you accept images, audio or video, check them for hidden instructions and manipulations, such as text in an image or noise designed to fool the model.
What good looks like
- Text inside images and speech in audio is extracted and run through the same checks as typed input.
- Files are checked for hidden data and known attack patterns.
- Suspicious files are blocked or quarantined.
How teams usually do it
- Optical character recognition and speech-to-text before the injection and content checks.
- Re-encoding or resizing images to break many pixel-level perturbations.
Evidence an auditor would accept
- The processing steps for each media type.
- Test results with images containing hidden instructions.
Common mistakes
- Screening text input carefully while passing images straight to the model.
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.