KryptonGRC Academy
The lead class
Learn about AI auditing and mapping to frameworks.
What you take in: a test run, an agent run or a production set — a record taken from live use. You read it whole — never one turn alone.
Each criterion is read against the book. It stands one of three ways — never a fourth.
You make your call first. The reasoning opens after it — never before.
The finding you write lands in the register — the list where every finding stands — and carries into the report, and into your portfolio.
Step 1 · Take the record — the Auditor’s intake, the questions the Auditor asks first: four questions, in their order
What did you drop?
Work the class, then bring out what your work made. One tap generates it, from your own work: the register, the control matrix, the plan of action and the charts. Together they are your portfolio — the class's ending, and its proof.
One door, not a sequence. Nothing here is typed in over your work — every line below is generated by it, and all of it carries into your report.
Every check you work lands here as a row: its control, what the control maps to and its remediation. The register carries into your report.
| Worked item | Control | Maps to | Remediation |
|---|---|---|---|
| Step 2 — a check on testing, read against the framework | AI test, evaluation, verification & validation programme | NIST AI RMF: MEASURE 1.1 · MEASURE 1.2 · MEASURE 2.1 · MEASURE 2.13 · MEASURE 2.2 · MEASURE 2.3 · MEASURE 2.5 · MEASURE 4.1 · MEASURE 4.2 · ISO/IEC 42001: 9.1 · A.6.2.4 · MITRE ATLAS: covered another way, through Secure development and testing — AML.T0016 · AML.T0119 | Every gap the verdict leaves takes its own line in the plan of action below |
| Step 4 — a failure found in production | A named gap — no control is named for this item in the class | No seat named | A named retest. It closes on proof: a later record shows the retest run, and passed |
The controls that speak to testing, read against the class's three frameworks. A seat stands covered where the book carries it; where the book covers the point another way, the way stands with it.
| Control | MITRE ATLAS | NIST AI RMF 1.0 | ISO/IEC 42001:2023 |
|---|---|---|---|
| AI test, evaluation, verification & validation programme | Accounted — covered another way: Secure development and testing carries AML.T0016 · AML.T0119 | MEASURE 1.1 · MEASURE 1.2 · MEASURE 2.1 · MEASURE 2.13 · MEASURE 2.2 · MEASURE 2.3 · MEASURE 2.5 · MEASURE 4.1 · MEASURE 4.2 Covered | 9.1 · A.6.2.4 Covered |
| Secure development and testing | AML.T0016 · AML.T0119 Covered | GOVERN 4.3 · MAP 1.6 Covered | A.6.1.2 · A.6.1.3 · A.6.2.2 · A.6.2.3 · A.6.2.5 Covered |
The class's gaps and failures, as a plan. A seat with no owner or date yet reads Not assigned.
| Item | Milestone | Owner | Date |
|---|---|---|---|
| The verdict's named gap — one criterion stands as a named gap (Step 3) | Its own line in the plan of action; closes on proof | Not assigned | Not assigned |
| A failure found in production | The named retest — closes on proof: a later record shows the retest run, and passed | Not assigned | Not assigned |
| The production failure's control seat stands unnamed in the class | Close on proof: a control stands named for the row | Not assigned | Not assigned |
This class's standings, in the snapshot's form — and its matrix, drawn.
This class stands mostly covered, with one seat accounted and no gaps.
Findings in the plan of action: 3
| Control | ATLAS | AI RMF | 42001 |
|---|---|---|---|
| AI test & evaluation programme | Accounted | Covered | Covered |
| Secure development & testing | Covered | Covered | Covered |
Gold — covered · Navy — accounted · Dashed — a named gap.
Self-paced, like every course in the Academy. The path ends in your portfolio.