How to meet it
OWASP AISVS v1.01
Training Data Integrity & Traceability · Training Data Quality and Security Assurance
Remove disallowed content before training
Verify that disallowed content is detected and removed before training.
OWASP AISVS v1.01, requirement 1.3.4
In plain English
Find and remove content you do not want the model to learn from, such as abuse material, hate content or leaked secrets, before training starts.
What good looks like
- There is a written list of disallowed content categories.
- Automated filters run over new training data before it is used.
- Removed items are logged so the filter can be checked.
How teams usually do it
- Content classifiers and hash-matching against known prohibited material.
- Secret scanners run over code and text data.
Evidence an auditor would accept
- The disallowed content policy.
- Filter configuration and counts of removed items per run.
Common mistakes
- Filtering the base dataset but not data added later for fine-tuning.
- No secret scanning, so keys and passwords end up in 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.