Provenance checks on retrieval and training data
Documents the agent retrieves or is trained on come from known sources and are checked for tampering.
Layer L3 Application · Owner Deployer · Family Supply chain
Effective
Not effective
Not effective (bypassed)
Did not exist
By risk
| Risk | Incidents where the control was not effective or did not exist | Incidents where the control was effective | Incidents where sources don't mention the control |
|---|---|---|---|
| Supply-chain compromise | 5 | ||
| Memory and data poisoning |
Why it was not effective
No incident yet records this control as not effective.
Incidents where it was not effective or did not exist per month
No rise: 0 in Oct 2026, against an average of 0 over the 3 months before.
19 of 183 incidents in Oct 2026 have been read for controls.
Added in reported fixes, and what followed
No reported fix names this control yet.
Frameworks
| Standard | Clause | Requirement |
|---|---|---|
| OWASP LLM | LLM04 | To verify when the text is stored |
| MITRE ATLAS | AML.M0007 | To verify when the text is stored |