The learning, on its own name

KryptonGRC Academy

AI Auditing

Learn about AI auditing and mapping to frameworks

  • Agent runs, read whole — every turn, every tool call
  • Eval records — the dataset, the points it was judged on, the scores
  • One agent, four parts: planning, sub-agents, files, instructions

Open the AI auditing class

AI Framework Courses

The AI frameworks first — one course per framework, from the book

Open the first course — ISO/IEC 42001:2023

Stage 1 · Test the policy

Policy Tester — four lessons

  • Storage, public and unencrypted
  • The sign-in door, open to the whole internet
  • One access key that opens everything
  • The most powerful user, signing in with a password alone

Stage 2 · Judge the evidence

Practice Lab — six exercises, and the worked example set whole

  • Access reviews
  • Monitoring activity
  • Privileged access and MFA
  • Encryption at rest
  • Backup and restore
  • Brightline Mutual — the claims helper nobody registered

Stage 3 · Run the audit

Projects — two full audits

  • Apex Financial Services
  • MedCore Health Services

Stage 4 · Prove it

The portfolio — where the path ends

  • Your full report, and the record of every call you made

Framework Courses

Then every other framework in the book takes its own course, at its own grain — 104 in all.

Start where it makes sense. Every course is self-paced, and the path ends in your portfolio.

New here? Read the Guide — how the Academy works, in plain words.

Not sure where to start? Answer three questions

Start-here sets

Session workspace

Every number on this board comes from your own actions.

Status · Ready — nothing tested yet Period · This visit Suites · Policy Tester · Practice Lab

Dashboard

This visit · live session

Your work

ASNone loaded
0
Total Assets
RKAll clear
0
Active Risks
Every row raised on the register, rated or not
CTNothing judged
0
Controls
Judged across every surface that judges
OIAll clear
0
Open Issues
Actions awaiting a rating or a treatment — risks accepted or monitored are not issues
SENone yet
0
Sealed Reports

Health score

Health score
Nothing tested yet

Run an assignment or judge a scenario and the score wakes up.

Risk distribution

0RATED
High0
Medium0
Low0

No risks yet. The first one is raised the moment a policy test fails. That is where this board begins.

Control Status0 of 18 judged

No control has been tested yet.

Suite progress

Policy Tester0 / 4
Practice Lab0 / 6
Trilogy Lab0 / 9
ISO 42001 Board0 / 9

Control effectiveness

—

Nothing judged yet. The first verdict on any surface wakes this panel.

Three surfaces judge work in this Academy. Their verdicts gather here.

Treatment coverage

—

No risks raised yet. Coverage waits for the first finding.

Inherent and residual risk

InherentThe exposure as raised — every rated risk at the rating it was given.
ResidualWhat still stands untreated — a seat empties only when its cause is fixed.

Recent Findings0 standing

No findings yet. A policy test that failed, a restricted asset or an evidence gap raises the first one here — and a call of yours standing against the auditor is recorded here too, as yours.

Modules and frameworks

The Academy modules and the frameworks the work stands on, in one place. Every line opens the room where its work happens.

Also on the framework board: COBIT 2019, held as a benchmark.

No systems loaded — load a pack in AI Data Readiness Fixes applied: 0
Session detail
Calls agreed with the auditor0 / 0
Rules progress0 / 0
Agreed progress0 / 0

Program

A working GRC program is a chain: risk, controls, evidence, reporting.

Knowledge checks
Krypton GRC
Terafram · Audit software™
Audit Report — this session
The print dialog can save this report as a PDF.
1

Executive summary

2

Objective & scope

This workspace tests cloud policy rules and trains audit judgement on evidence packs. It is a teaching and demonstration workspace, not an audit of any real organisation. Nothing here examines live systems, a client system or a named company. The assignments use published sample configurations. The scenarios use written evidence packs prepared for training.

3

Methodology & audit approach

The work runs on CISA audit method with PMP execution discipline. Working to a method is not the same as holding its certification, and finishing work in this Academy awards no certification of any kind. Evidence comes before conclusions. Every policy verdict traces back to the output printed with the run. Every lab conclusion traces back to the auditor's stated reasoning on the evidence pack. A result you cannot trace is a result you cannot defend, so this workspace shows its working for every call it records.

Operational Audit Logic

CISA — Audit · PMP — Audit Execution · Security AI — AI Governance

Full closed-loop audit software, from beginning to end.

4

Provisions

One row per item across both suites. Status and conclusion reflect your latest action. Review jumps to the item in its room.

ItemSuiteMethodStatusConclusionStanding
5

AI Data Readiness

6

Seal the session

Seal this report into the Krypton Seal evidence chain. The browser hashes the summary above with SHA-256 (the Secure Hash Algorithm — a standard method that turns any text into a fixed fingerprint, so changing one word changes the fingerprint) and the hash — never the report itself — is entered into the chain. The seal is tamper-evident: change one word of the summary afterwards and the hash no longer matches the entry. If the seal service does not answer, the report stands unsealed and this section says so.

Built by Terry Afram-Kumi · CISA · PMP · MSc · ISO/IEC 27001:2022 Lead Auditor · AI Security & Governance Certified — Founder, Terafram

Portfolio

Everything you earn in this Academy gathers here — sealed evidence, not souvenirs.

The gathered record

One printable document holding all four sections — the stamps with their dates, the lab reports, the session summary and the sealed entries — under the house story and open to the Seal.

Projects

One project at a time — work it through the whole chain: policy, controls, risk, remediation, report.

Policy Tester

AssignmentFactsStandingAction

The policy tester is adapted from the open-source GRC Playground by Ashley Pearce (MIT). Scenarios and page by Terafram.

Practice Lab

Pick a scenario. Read the situation and the evidence, then commit to a call.

Mode
ScenarioControlYour callAuditor's callStandingAction

WORKED EXAMPLE · ILLUSTRATIVE TRAINING DATA

Privileged Access Review

The document's worked example, built whole: a fictional financial-services company, a privileged account review and the five decisions an auditor works through — findings, conclusion, framework, requirement and remediation. Its result stands here and nowhere else: no register row, no dashboard count.

Not attempted yet this visit.

SOX

A Krypton Audit Academy training register

The SOX (Sarbanes-Oxley) and ICFR (internal control over financial reporting) control register. Five controls on the systems of a labelled scenario company — the whole of one fictional company in this exercise: its systems, its records and its financial close — worked for training.

This register teaches method. It issues nothing. Krypton does not issue SOX certification, attestation or an audit opinion — not for the learner, not for these systems, not for anyone. We don't issue certification but we give guidance and assurance.

A control is one of the checks the company runs in its own work so that its accounts can be trusted, and each line below is one of those checks. One control stands under each component of the COSO framework (COSO is the Committee of Sponsoring Organizations of the Treadway Commission, and its framework is the standard way internal controls are organised), in the grammar of a SOX Section 302 and Section 404 register: a reference, a type, the assertion it serves, an effectiveness standing and a walkthrough date. Preventive means the control stops the problem before it happens. Detective means it finds the problem after it has happened, so it can be put right. The assertion is the claim about the accounts that the control backs — that a recorded sale really happened, that nothing is missing, that the company holds the rights it says it holds. A walkthrough is following one item through the process with the people who run it, to watch the control work as it really works. Every row starts Not tested and every Walkthrough cell stands empty. A walked-through control stands Effective when it did its job on the day it was followed, Deficient when it did not do its job, and Not enough evidence when the walkthrough could not gather enough to judge it — that standing is provisional, neither passed nor failed, until the evidence can settle it. A date lands in that cell only on the day a walkthrough is really performed in this room, with the effectiveness standing earned in that same walkthrough against the auditor's stamped conclusion. Nothing on this register is seeded and no standing is ever set by hand.

Register grammar adopted from the open GRC Practice Lab by John Bommeraveni Joseph. The room, its scenario systems and its judgement are Krypton's.

Frameworks

One table: every framework, and where you stand against it.

Book 0.7.75 · 104 frameworks · 180 controls — every standing on this page reads from the live book.

FrameworkTypeReachAction
The standard library — the twelve
ISO/IEC 27001:2022CertificationFull Scan — reaches 85 of the 180 controls
NIST CSF 2.0BenchmarkFull Scan — reaches 39 of the 180 controls
SOC 2 (Trust Services Criteria)AttestationFull Scan — reaches 68 of the 180 controls
PCI DSS (Payment Card Industry Data Security Standard) v4.0CertificationFull Scan — reaches 42 of the 180 controls · template in Policy Pack Part II
HIPAA (Health Insurance Portability and Accountability Act) Security RuleRegulatoryFull Scan — reaches 29 of the 180 controls · templates in Policy Pack Part II
GDPR (General Data Protection Regulation, EU 2016/679)RegulatoryFull Scan — reaches 54 of the 180 controls
CIS Controls v8BenchmarkFull Scan — reaches 35 of the 180 controls
HITRUST (Health Information Trust Alliance) CSFCertificationFull Scan — reaches 21 of the 180 controls
NIST SP 800-53 (Rev. 5)RegulatoryFull Scan — reaches 96 of the 180 controls
CSA Cloud Controls Matrix v4.1BenchmarkFull Scan — reaches 71 of the 180 controls
ISO/IEC 42001:2023CertificationFull Scan — reaches 33 of the 180 controls · ISO 42001 Board and Academy
COBIT (Control Objectives for Information and Related Technologies) 2019BenchmarkFull Scan — reaches 44 of the 180 controls
Krypton's additional twelve
NIST SP 800-171 (Rev. 3)RegulatoryFull Scan — reaches 58 of the 180 controls
CMMC (Cybersecurity Maturity Model Certification) 2.0 (Level 2)CertificationFull Scan — reaches 49 of the 180 controls
FedRAMP (Federal Risk and Authorization Management Program)RegulatoryFull Scan — reaches 29 of the 180 controls
NIST AI RMF 1.0BenchmarkFull Scan — reaches 29 of the 180 controls
SOC 1 (SSAE 18 / ISAE 3402)AttestationFull Scan — reaches 14 of the 180 controls
ISO 9001:2015CertificationFull Scan — reaches 28 of the 180 controls
OWASP Top 10:2025BenchmarkFull Scan — reaches 18 of the 180 controls
OWASP API Security Top 10:2023BenchmarkFull Scan — reaches 16 of the 180 controls
OWASP LLM (large language model) Top 10:2026BenchmarkFull Scan — reaches 17 of the 180 controls
OWASP CWEBenchmarkFull Scan — reaches 22 of the 180 controls
MITRE ATT&CK (Enterprise)BenchmarkFull Scan — reaches 29 of the 180 controls
MITRE ATLAS (AI Threat Matrix)BenchmarkFull Scan — reaches 31 of the 180 controls
FFIEC (IT Examination Handbooks)RegulatoryFull Scan — reaches 37 of the 180 controls
FERPA (34 CFR Part 99)RegulatoryFull Scan — reaches 11 of the 180 controls
OWASP Application Security Verification Standard (ASVS) 5.0BenchmarkFull Scan — reaches 26 of the 180 controls
CSA AI Controls Matrix (AICM) v1.1BenchmarkFull Scan — reaches 23 of the 180 controls
ISO/IEC 23894:2023 (AI Risk Management)BenchmarkFull Scan — reaches 7 of the 180 controls
NIST SP 800-218 (SSDF v1.1)BenchmarkFull Scan — reaches 18 of the 180 controls
NIST Privacy Framework v1.0BenchmarkFull Scan — reaches 53 of the 180 controls
CCPA/CPRA (California Consumer Privacy Act)RegulatoryFull Scan — reaches 9 of the 180 controls
SWIFT CSCF (Customer Security Controls Framework) v2026BenchmarkFull Scan — reaches 11 of the 180 controls
COSO Internal Control - Integrated Framework (2013)BenchmarkFull Scan — reaches 14 of the 180 controls
IEC 62443-3-3 (Industrial Automation and Control Systems Security)BenchmarkFull Scan — reaches 11 of the 180 controls
NERC CIP (Critical Infrastructure Protection)RegulatoryFull Scan — reaches 28 of the 180 controls
EU Cyber Resilience Act (Regulation (EU) 2024/2847)RegulatoryFull Scan — reaches 14 of the 180 controls
ISO/IEC 20000-1:2018 (Service Management)CertificationFull Scan — reaches 20 of the 180 controls
IEEE Std 2863-2026 (Organizational Governance of Artificial Intelligence)BenchmarkFull Scan — reaches 23 of the 180 controls
ASD Essential Eight (ACSC Essential Eight Maturity Model)BenchmarkFull Scan — reaches 8 of the 180 controls
SLSA (Supply-chain Levels for Software Artifacts) v1.2 - Build TrackBenchmarkFull Scan — reaches 5 of the 180 controls
NIST SP 800-53A (Assessing Security and Privacy Controls in Information Systems and Organizations)BenchmarkFull Scan — reaches 26 of the 180 controls
DOE Cybersecurity Capability Maturity Model (C2M2) v2.1BenchmarkFull Scan — reaches 46 of the 180 controls
MAS TRM (Monetary Authority of Singapore Technology Risk Management Guidelines, January 2021)RegulatoryFull Scan — reaches 30 of the 180 controls
OWASP SAMM (Software Assurance Maturity Model) v2BenchmarkFull Scan — reaches 17 of the 180 controls
NCSC Cyber Assessment Framework (CAF) v4.0BenchmarkFull Scan — reaches 30 of the 180 controls
ISO/IEC 27031:2025 (ICT Readiness for Business Continuity)BenchmarkFull Scan — reaches 11 of the 180 controls
Colorado AI Act (SB 24-205)RegulatoryFull Scan — reaches 10 of the 180 controls
NIST SP 800-218A (SSDF Community Profile for Generative AI and Dual-Use Foundation Models)BenchmarkFull Scan — reaches 11 of the 180 controls
ISO/IEC 42005:2025 (AI System Impact Assessment)BenchmarkFull Scan — reaches 8 of the 180 controls
BSI C5:2020 (Cloud Computing Compliance Criteria Catalogue)AttestationFull Scan — reaches 62 of the 180 controls
TISAX (VDA ISA 6.0.3)CertificationFull Scan — reaches 26 of the 180 controls
Brazil LGPD (Lei nº 13.709/2018)RegulatoryFull Scan — reaches 37 of the 180 controls
Ghana Data Protection Act, 2012 (Act 843)RegulatoryFull Scan — reaches 30 of the 180 controls
Kenya Data Protection Act, 2019 (No. 24 of 2019)RegulatoryFull Scan — reaches 33 of the 180 controls
South Africa POPIA (Protection of Personal Information Act, 2013)RegulatoryFull Scan — reaches 32 of the 180 controls
ETSI EN 303 645 V3.1.3 (Cyber Security for Consumer Internet of Things: Baseline Requirements)BenchmarkFull Scan — reaches 26 of the 180 controls
IEC 62443-4-1:2018 (Secure Product Development Lifecycle Requirements)BenchmarkFull Scan — reaches 17 of the 180 controls
ISO/IEC 27005:2022 (Guidance on Managing Information Security Risks)BenchmarkFull Scan — reaches 11 of the 180 controls
India DPDP Act 2023 (Digital Personal Data Protection Act, 2023)RegulatoryFull Scan — reaches 15 of the 180 controls
EU AI Act (Regulation (EU) 2024/1689, as amended)RegulatoryFull Scan — reaches 28 of the 180 controls
BSI IT-Grundschutz-Kompendium (Edition 2023)BenchmarkFull Scan — reaches 48 of the 180 controls
Spain ENS (Esquema Nacional de Seguridad, Real Decreto 311/2022)CertificationFull Scan — reaches 50 of the 180 controls
EBA Guidelines on ICT and Security Risk Management (EBA/GL/2019/04, as amended)RegulatoryFull Scan — reaches 3 of the 180 controls
eIDAS (Regulation (EU) No 910/2014, as amended by Regulation (EU) 2024/1183)RegulatoryFull Scan — reaches 14 of the 180 controls
PSD2 Security Requirements and SCA RTS (Directive (EU) 2015/2366; Delegated Regulation (EU) 2018/389)RegulatoryFull Scan — reaches 20 of the 180 controls
UK GDPR and Data Protection Act 2018RegulatoryFull Scan — reaches 10 of the 180 controls
UK PSTI Act 2022 (Product Security and Telecommunications Infrastructure Act, Part 1)RegulatoryFull Scan — reaches 9 of the 180 controls
NCSC Cloud Security PrinciplesBenchmarkFull Scan — reaches 14 of the 180 controls
UK Cyber Essentials (Requirements for IT Infrastructure v3.3)CertificationFull Scan — reaches 15 of the 180 controls
Personal Information Protection and Electronic Documents Act (PIPEDA), Part 1 and Schedule 1RegulatoryFull Scan — reaches 27 of the 180 controls
CCCS Baseline Cyber Security Controls for Small and Medium OrganizationsBenchmarkFull Scan — reaches 27 of the 180 controls
Act respecting the protection of personal information in the private sector (Québec), as amended by Law 25RegulatoryFull Scan — reaches 35 of the 180 controls
ITSG-33 IT Security Risk Management: A Lifecycle ApproachBenchmarkFull Scan — reaches 25 of the 180 controls
ISO/IEC 15408:2022 (Common Criteria - Evaluation Criteria for IT Security)CertificationFull Scan — reaches 40 of the 180 controls
Australian Government Information Security Manual (ISM)BenchmarkFull Scan — reaches 57 of the 180 controls
OWASP Mobile Application Security Verification Standard (MASVS) 2.1.0BenchmarkFull Scan — reaches 17 of the 180 controls
StateRAMP (operating as GovRAMP) Security RequirementsCertificationFull Scan — reaches 13 of the 180 controls
Saudi NCA Essential Cybersecurity Controls (ECC-2:2024)RegulatoryFull Scan — reaches 46 of the 180 controls
India CERT-In Directions under Section 70B (No. 20(3)/2022-CERT-In, 28 April 2022)RegulatoryFull Scan — reaches 9 of the 180 controls
Singapore Cybersecurity Code of Practice for Critical Information Infrastructure (2026)RegulatoryFull Scan — reaches 49 of the 180 controls
ISO 31000:2018 (Risk management — Guidelines)BenchmarkFull Scan — reaches 9 of the 180 controls
Taught in the Academy
COSOInternal control frameworkThe SOX Register — five components and five controls
A framework score is an auditor's view of where you stand against it. It is not a certification. Terafram does not issue certifications — we give guidance and assurance.

Risk Register

Findings become risks here.

Likelihood × impact

Rows are likelihood, columns are impact. Likelihood is how likely the harm is to happen, scored Low (1), Medium (2) or High (3). Impact is how much harm it does if it happens, scored on the same scale. The number in a cell is how many rated risks sit there.

No risks raised yet. Fail a policy lesson, judge a scenario Not operating or load systems whose restricted data feeds AI training or retrieval (RAG) and the finding lands here as a risk.

Policy Pack

The register of both sets — the fourteen house policies Terafram runs on, and the client templates issued under guidance.

Part I — 14 Terafram internal · Part II — 6 client adoption templates
PolicyCategoryPartOwnerStatusThis visitAction

The texts are issued inside a guided adoption, not downloaded here — a template in the wrong scope is paper compliance.

AI Adoption Policy Client template Approved for issue
Approval progress
  • ✓Drafted
  • ✓Reviewed
  • ✓Approved
  • ✓Published

Approved for issue by the founder on 25 October 2026 and published in the pack the same day.

Version history
VersionStatusDate
1.0Approved for issue25 October 2026
1.1Credential line completed to the full standing set25 October 2026
Auditor's note
We don't issue certification but we give guidance and assurance — an auditor's independent view of where you stand.
Terafram — Controlled Document · Client Adoption Template

AI Adoption Policy

Owner: AI Governance Lead
Version: 1.1
Status: Client adoption template — approved for issue, 25 October 2026
Issued: 25 October 2026
Review: At least annually
1. Purpose

This policy governs how the adopting organisation's workforce adopts and uses artificial intelligence tools. It brings AI use into the open, classifies every tool before it is used and fixes the rules for the data a tool may touch, the review its output must pass and the limits its agents must work inside.

2. Scope

This policy applies to every use of an AI tool, AI feature or AI agent by the adopting organisation's workforce, including employees, contractors and temporary staff, at any location and on any device used for work. It covers tools bought by the organisation, tools provided inside other software and tools adopted by individuals on their own.

3. Policy Statements
Approved Approved with masking Restricted

Twelve auditable statements govern classification before use, data gating, human review of output, agent limits and AI incidents — issued in full inside a guided adoption.

7. Related Controls

References for this policy:

  • ISO/IEC 42001 — Artificial intelligence management systems. This template is aligned to the principles of ISO/IEC 42001 as a statement of design intent only. Adoption of this template is not a claim of certification to or conformity with ISO/IEC 42001.
  • POL-013 — AI Acceptable Use & Governance Policy: the internal twin of this template, which Terafram itself operates under Part I of this pack.

Issued by: Terry Afram-Kumi — 25 October 2026

Adopted by (client): ______ · Date: ______

How a guided adoption runs
1 · Scenario first Your situation, told in your facts. The worked example shows the shape your adoption follows.
2 · Scope agreed What the policy governs gets named: the data, the systems, the people. What sits outside it gets named too.
3 · Template issued The template leaves Terafram with your owners and your systems written in, ready for your leadership to read.
4 · Client adopts Your leadership signs. Adoption is the client's act; the policy becomes yours on that signature.
5 · Auditor verifies A CISA pass over the evidence: does the policy operate, not just exist?
We don't issue certification but we give guidance and assurance. Terafram is not a certification body. Guidance walks the adoption with you. Assurance is an auditor's independent view of where you stand.

Trilogy Lab

This room runs on one chain, in this order: Policy → Controls → Risk → Remediation → Report Generation.

StationManual partBeatChain linkStatus

The policies quoted in this room are the approved Terafram documents. The check is the same check the Policy Tester uses, read here in the Academy.

ISO 42001 Board

0/9 objectives tested
The management-system spine — Clauses 4 to 10
Clause 4Context
Clause 5Leadership
Clause 6Planning
Clause 7Support
Clause 8Operation
Clause 9Performance evaluation
Clause 10Improvement

The nine objectives below sit under this spine. The spine is the standard's shape alone — every state on the board comes from your work.

A row moves only when the work happens
CodeObjectiveControlsEvidenced byMethodStandingAction

ISO/IEC is the International Organization for Standardization and the International Electrotechnical Commission, and 42001 is their standard for AI management systems. Where a row's method says Walkthrough, it means following one item through the process with the people who run it, to watch the control work as it really works. Conclusions here are the lab's verdicts on session work — they are not a certification claim, for anyone, ever.

The board reads its tested rows from your runs, and your runs rest on this device — leave the page and its rows stand as you left them when you come back. Your Academy stamps are saved on this device and wait for you below.

Modules completed 0/12

Evidence & Results

Every policy run and every lab judgement stands here, each with its evidence.

Nothing here yet. Your first policy verdict and your first lab judgement land here the moment they happen.

AI Data Readiness

Judged live — declared assets only

Which assets are safe to feed into AI — and what must be masked or kept out.

Load scenario

Three doors from here

The main site

The KryptonGRC Academy is the workshop. The main site carries the full story: how Krypton works, the frameworks it covers and the pricing.

Visit kryptongrc.com

Your own systems, tested for real

The Playground judges sample files. The workspace is where this work grows into the full audit — assessments across 103 frameworks, cloud evidence collected and sealed, agents tested before they act. Express is free. Your first full scan is free.

Enter the Krypton GRC workspace

Talk it through first

Prefer to talk it through first? Book a free consultation — thirty minutes with me, not a sales rep.

Book a free consultation

Why I give this away

I am Terry Afram-Kumi. I audit for a living and I built Krypton GRC so a verdict never depends on who is asking. The Playground is the front porch: small, free and honest about what it proves. If it teaches you the pattern, it has done its job — and if you want the pattern run across all your systems, you know where the door is. Teaching with it? Run it in your classes and your videos — that is what it is for.