Recognises a control tested for design only
test: design — the matrix records that a control exists, not that it runs. Supervisors test operating effectiveness.
Can AI do this?
Not settled yet The benchmark tested this, but the models' answers do not settle yet whether they master it. 0 of 4 models in the current benchmark master it.
| Model | Tier | Probability of mastery | Status |
|---|---|---|---|
| gemini-4-argon | Flagship | 31% | Not shown |
| gemini-3.5-flash-lite | Low cost | 4% | Not shown |
| gpt-6-luna | Low cost | 4% | Not shown |
| claude-haiku-5-5 | Low cost | 32% | Not shown |
Mastered means a probability of at least 90% that the model has this knobit, estimated from its graded answers. “Not shown” means the answers do not settle it either way. How the models are tested.
What it looks like in practice
Expected result: The control and the operating evidence needed
How it is assessed: Flags the deficient case and passes the acceptable one. Deficient: CT-004 Loan exposure reconciliation, test: design, evidence "Reconciliation procedure document" Acceptable: test: both; evidence is each month's signed reconciliation with its breaks and their resolution
Also known as: Design tested, operation never
Related knobits
Builds on
All 21 knobits of Map BCBS 239 to controls.
Source: Review checklist 04-regulatory-mapping, row 6 (ECB; DCAM)