Method
RAG Contamination Threat Model
Maps contamination paths from source material through retrieval into model output and downstream action.
Field scoring contract
| Field | Type | Role | Direction | Scored |
|---|---|---|---|---|
| Source trust | select | risk_driver | higher_is_worse | yes |
| Ingestion governance | select | risk_driver | higher_is_worse | yes |
| Retrieval access control | select | risk_driver | higher_is_worse | yes |
| Sensitive content | select | risk_driver | higher_is_worse | yes |
| Citation validation | select | risk_driver | higher_is_worse | yes |
| Downstream action | select | risk_driver | higher_is_worse | yes |
Limits
- Preliminary output
- Human review required
- Not certification
- Risk/readiness separated where implemented
- Context text is not averaged into numeric scores.
- Outputs require source/evidence review before decisions.