KryptonGRC Academy

The first framework course

ISO/IEC 42001:2023

Learn about ISO/IEC 42001:2023 and how its controls map.

Phase 1 · Foundations

ISO 42001ISO 27001AI RMF

What an AI management system is, and where 42001 sits beside 27001 and the AI Risk Management Framework.

Phase 2 · Context — Clause 4

The AI systemThe organisationPartiesParties

Scope comes before controls: the organisation, its interested parties and the boundary of its AI management system.

Phase 3 · Leadership & Planning — Clauses 5–6

PolicyRolesRiskmethod

Policy, roles and the risk method that turns AI use into governed work.

Phase 4 · Support & Operation — Clauses 7–8

SkillsDocsLife cycleData

Skills, documents, the system life cycle and data: the working machinery.

Phase 5 · Evaluation & Improvement — Clauses 9–10

MeasureAuditReviewImprove

Measure, audit, review and improve — and the report that proves it.

On the book: ISO/IEC 42001:2023 · item grain · catalogue 61 · covered 61 · accounted 0 · gaps named: none.

The class at work

Module 5.1 · Use of AI systems, judged — four rulings at Station 5

Question 1Control A.9 · Use of AI systems

An agent proposes. The document rules. An AI agent issues a refund to a customer, above the limit its instructions set.

Which statement covers this action? Tap the statement.

Client dataClient data, client documents and confidential Terafram information must never be entered into an AI service unless the contract with the provider prohibits the provider from using that data to train models and the client engagement terms permit the use.
Human review of outputAI-generated output must be reviewed by a qualified person before it is relied on, sent to a client or used in an audit, assessment or report. The reviewer is accountable for the output as if the reviewer had produced it personally.
No autonomous commitmentsAI systems and agents must not make commitments to clients, issue audit conclusions or modify client data without the approval of a named human.
Agent oversightAI agents operated by or for Terafram must run with least-privilege credentials in accordance with the access control policy, must have their actions logged in accordance with the logging and monitoring policy and must be capable of being stopped immediately by a named human operator.

The AI Acceptable Use & Governance Policy in full stands behind this question — one tap away, as reading material.

BackDecision 1 of 4Continue
Question 2Control A.9 · Use of AI systems

An agent proposes. The document rules. An AI agent issues a refund to a customer, above the limit its instructions set.

What does the statement rule? Tap the ruling.

AllowedAs it stands
Human loopA named human decides
Not allowedThe document stops it
BackDecision 2 of 4Continue

Your portfolio — one tap

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.

Self-paced, like every course in the Academy. The path ends in your portfolio.

Onward — the AI auditing class  ·  Back to the Academy