How to meet it
OWASP AISVS v1.01
Access Control & Identity for AI Components & Users · AI Resource Authorization & Classification
Fetch sensitive data, don't train it in
Verify that sensitive data is retrieved via retrieval pipelines (e.g., RAG queries, embedding lookups) to prevent permanent storage in models.
OWASP AISVS v1.01, requirement 5.2.3
In plain English
Keep sensitive data out of model training and fetch it at answer time through retrieval instead. Data trained into a model cannot be deleted or permission-checked later.
What good looks like
- Sensitive data is excluded from training and fine-tuning sets.
- The assistant reaches sensitive data through retrieval with access checks.
- Deleting the source document removes it from future answers.
How teams usually do it
- Retrieval-augmented generation over permissioned stores for anything confidential.
- A data classification check before data enters a training set.
Evidence an auditor would accept
- The policy on what may be used for training.
- Training dataset reviews showing sensitive data was excluded.
Common mistakes
- Fine-tuning on support tickets that contain customer personal data.
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.