ISO 42001, implemented.

Aegentra scopes and implements an ISO/IEC 42001 AI management system around your real AI estate. Roles, applicable controls, evidence and timeline are agreed before work starts; any internal audit is separately scoped and conflict-checked.

What ISO/IEC 42001:2023 is

ISO/IEC 42001:2023 is the world's first international management system standard for artificial intelligence, published by ISO and IEC in December 2023. It does not tell you how to build a particular model; it tells you how to govern AI as a portfolio — how to inventory the AI systems you develop or deploy, assess their impact on people and the organisation, embed human oversight, control the data that trains and feeds them, and prove the whole system is monitored and improved over time. The output is an AI Management System (AIMS): a living governance layer a certification body can audit and that a procurement team, a regulator, or a board can trust. Structurally it uses the same Annex SL High-Level Structure as ISO 27001, so if you already run an information security management system, the shape is immediately familiar — the difference is a new set of AI-specific controls.

• First international AI management standard — ISO/IEC 42001:2023, published December 2023
• Governs an AI Management System (AIMS), not a single model
• Certifiable by a JAS-ANZ-accredited body; certificate valid three years
• Built on Annex SL — the same governance structure as ISO 27001
• Applies whether you develop, provide, or use AI systems

Who needs ISO 42001 in Australia

ISO 42001 is voluntary in Australia, but the pressure to hold it is no longer theoretical. Three forces are driving demand. Enterprise procurement now asks AI-specific questions in vendor security questionnaires — buyers want proof that the AI inside your product is trained on lawful data, monitored for drift, and supervised by humans where decisions matter. The EU AI Act, in force since August 2024 and phasing in through 2027, makes AI governance a legal condition of selling into Europe, and ISO 42001 is the most defensible evidence base a provider or deployer can point to. And Australian government AI policy — the Voluntary AI Safety Standard, the DTA's policy for the responsible use of AI in government, and state assurance frameworks — increasingly flows to suppliers through contract, making a certifiable AIMS the efficient way to answer a tender. If you build AI, embed a third-party model, or ship AI features to customers, ISO 42001 is the standard those audiences will ask you to evidence.

• AI developers and deployers — anyone who builds, embeds, or ships AI features
• SaaS companies selling into the EU under the EU AI Act (in force Aug 2024, phased to 2027)
• Government suppliers inheriting AI-governance obligations through contract
• Regulated sectors — healthcare (TGA SaMD), finance (APRA CPS 230/234), legal
• Any organisation now facing AI questions in some enterprise procurement processes questionnaires

The AIMS structure: Clauses 4–10 and 38 Annex A controls

An ISO 42001 AIMS is defined by two parts. The management-system clauses (Clauses 4 to 10) set the governance loop — context, leadership, planning and AI risk assessment, support and competence, operation, performance evaluation, and continual improvement — identical in shape to ISO 27001. Annex A then adds 38 AI-specific controls, grouped under nine control objectives (A.2 to A.10): policies related to AI, internal organisation, resources for AI systems, assessing the impact of AI systems, the AI system life cycle, data for AI systems, information for interested parties, use of AI systems, and third-party and customer relationships. As with ISO 27001, you select applicable controls in a Statement of Applicability and justify any exclusions. The AI system impact assessment (AIIA) is the centre of gravity: it forces you to reason explicitly about bias, accuracy, robustness, transparency, privacy, safety, and accountability for every system in scope, and to document how each risk is treated.

• Clauses 4–10 — the Annex SL governance loop, from context to improvement
• Annex A — 38 controls across 9 objectives (A.2 policies to A.10 third-party relationships)
• Statement of Applicability — applicable controls selected, exclusions justified
• AI system impact assessment (AIIA) — bias, accuracy, transparency, privacy, safety
• Supporting annexes — B (implementation guidance), C (risk sources), D (sector use)

How Aegentra implements ISO 42001 inside Microsoft 365

Most AI governance work stops at a policy binder. We implement. A named Aegentra lead starts from your real AI estate — Microsoft 365 Copilot, Azure OpenAI, embedded model APIs, and any AI features inside your own product — and builds the AIMS from what is actually running, not a generic template. Because the majority of Australian SMB AI use now flows through the Microsoft 365 and Azure tenant, we configure the governance controls where the AI lives: Copilot data-access boundaries, sensitivity labelling for model inputs and outputs, DLP and audit logging around AI workflows, conditional access, and supplier controls for the model providers you depend on. The Aeges risk engine then runs against your tenant and produces live, on-demand evidence for each applicable control — so when the certification body asks for proof, you show a current system state, not a screenshot from three months ago. You end up with a working AIMS and a reproducible evidence pack, embedded in the environment your team already uses every day.

• We build the AIMS from your real AI estate — Copilot, Azure OpenAI, embedded APIs
• Governance controls configured where AI runs: your Microsoft 365 and Azure tenant
• Traceable evidence for applicable controls in scope via the Aeges risk engine
• Named accountability — the people who scope it are the people who build it
• The deliverable is a working AIMS and evidence pack, not a document binder

ISO 42001 vs ISO 27001

ISO 42001 is not a replacement for ISO 27001 — the two are complementary and share the same Annex SL backbone. ISO 27001 governs information security across the whole organisation; ISO 42001 governs the AI subset specifically, adding controls for the model life cycle, bias and fairness, transparency to the people AI decisions affect, human oversight, and AI-specific supplier obligations. If you already operate an ISO 27001 ISMS, roughly 60% of the framework — context, leadership, risk methodology, internal audit, management review — is reusable, which is why we run them as a single integrated management system rather than two parallel programmes. Neither standard is a prerequisite for the other, but most Australian AI-led companies implement ISO 27001 first because the wider buyer base understands it, then add ISO 42001 within six to twelve months. Running both under one governance loop, one internal audit, and one evidence pipeline is the sustainable target.

• ISO 27001 = information security org-wide; ISO 42001 = AI governance specifically
• Shared Annex SL structure — an existing ISMS reuses roughly 60% of the framework
• ISO 42001 adds model-lifecycle, bias, transparency, human-oversight and AI-supplier controls
• Neither is a prerequisite; most do ISO 27001 first, then ISO 42001 in 6–12 months
• Run them as one integrated ISMS + AIMS — a single audit and a single evidence pipeline

Timeline and cost in Australia

For an Australian organisation that already runs an ISO 27001 ISMS, a realistic readiness plan is around 12 weeks to audit-ready, covering a single AIMS scope over a defined portfolio of AI use cases. Starting greenfield — no ISMS, no model inventory, no AI register — is more realistically 16 to 22 weeks, most of the extra time spent inventorying current AI use and writing the governance documents an ISO 27001 implementer would already hold. Aegentra's implementation is fixed-scope and fixed-price, quoted after a short Week 0 discovery, so you know the number before any work begins — no hourly billing and no scope creep. Separately, the certification body's own Stage 1 and Stage 2 audit typically runs $10,000 to $22,000 AUD for an SMB, with lighter annual surveillance audits from around $2,500 in years one and two and a re-certification audit in year three. Book the certification body early: JAS-ANZ-recognised bodies are still scaling ISO 42001 audit capacity, so secure a Stage 2 slot in week one of readiness, not week eleven.

• About 12 weeks to audit-ready with an existing ISO 27001 ISMS
• 16–22 weeks greenfield (no ISMS, no model inventory)
• Aegentra implementation — fixed scope, fixed price, quoted after Week 0 discovery
• Certification body Stage 1 + Stage 2 audit — $10,000–$22,000 AUD (SMB, illustrative)
• Surveillance from ~$2,500/yr (years 1–2); re-certification in year 3
• Book the certification body early — ISO 42001 audit capacity is still scaling

The certification path

Aegentra implements and prepares the AIMS; the certificate itself is issued by a JAS-ANZ-accredited certification body — we never mark our own homework. The path runs the same way as ISO 27001. We inventory and scope every in-scope AI system, run the AI risk and impact assessments, design the AIMS and Statement of Applicability, implement and evidence each applicable control inside your tenant, then arrange a separately scoped internal audit by a party independent of implementation, followed by management review to close residual findings before the body arrives. The certification body performs a Stage 1 documentation review, then a Stage 2 effectiveness audit that samples evidence and interviews your engineers and product owners. The certificate is valid for three years, with surveillance audits in years one and two and full re-certification in year three. We sit in on both audit stages with you, and an optional retainer keeps the AIMS current as new models and use cases are deployed.

• Aegentra implements; a JAS-ANZ-accredited body certifies — independence preserved
• Stage 1 (documentation review) then Stage 2 (effectiveness audit, evidence sampling)
• Independent internal audit and management review before the body arrives
• Certificate valid three years — surveillance years 1–2, re-certification year 3
• Optional retainer keeps the AIMS current as new AI systems are deployed

Why Aegentra

AI governance attracts a lot of advice and very little implementation. The common offer is a template AI policy, a maturity spreadsheet, and a gap report your own team still has to action. Aegentra can log into your environment and build the AIMS by name, from the AI you actually run. A named lead carries the agreed scope from discovery through handover, and no policy binder that ages the day it is signed. Because we build the system where your AI lives and prove it with live Aeges evidence, your AIMS stays true to the tenant instead of drifting away from a static document set. That is the difference between holding a certificate and running a management system that survives the next surveillance audit.

• Implementation, not advice — the AIMS is built, configured, and evidenced
• Named accountability — scoped and delivered by the same team
• Built from your real AI estate inside Microsoft 365, not a generic template
• Live evidence via Aeges — a current system state, not a stale screenshot
• Fixed scope and fixed price, on-shore delivery, quoted after Week 0

Training the person who will own the AIMS is a different job from having it built: the PECB ISO 42001 Lead Implementer course teaches the method, and the ISO 42001 Lead Auditor course teaches how it gets tested. For what organisational certification costs and how long it takes, see the ISO 42001 certification guide for Australia. Clause 9.2 requires an internal audit by someone independent of the build — that is our ISO 42001 internal audit service.

The working documents are published free and ungated: a Statement of Applicability covering all 38 Annex A controls, an AI system inventory, a Clause 8.4 AI system impact assessment, and Australia’s 10 Voluntary AI Safety Standard guardrails mapped to ISO 42001 clauses.

ISO 42001 implementation — frequently asked questions

Is ISO 42001 mandatory in Australia?

No. ISO/IEC 42001 is voluntary in Australia, but it is the most widely recognised international standard for AI governance. Demand is driven by three things: some enterprise procurement processes may ask AI-specific questions in vendor security questionnaires; the EU AI Act (in force since August 2024, phasing in through 2027) makes AI governance a condition of selling into Europe; and Australian government AI policy — the Voluntary AI Safety Standard and the DTA's responsible-AI policy — increasingly flows to suppliers through contract. A certified AIMS is the efficient way to evidence all three.

Do I need ISO 27001 before ISO 42001?

No. Neither standard is a prerequisite for the other. But because ISO 42001 shares the Annex SL structure with ISO 27001, an existing ISMS reuses roughly 60% of the framework — context, leadership, risk methodology, internal audit and management review carry straight over. Most Australian AI-led companies implement ISO 27001 first (the wider buyer base understands it) and add ISO 42001 within six to twelve months, run as one integrated management system.

How long does ISO 42001 implementation take?

For an organisation that already runs an ISO 27001 ISMS, plan on around 12 weeks to audit-ready for a single AIMS scope. Starting greenfield — no ISMS, no model inventory, no AI register — is more realistically 16 to 22 weeks. The practical bottleneck is booking the certification body: JAS-ANZ-recognised bodies are still scaling ISO 42001 capacity, so secure a Stage 2 slot early rather than at the end of readiness.

What does ISO 42001 cost in Australia?

Aegentra's implementation is fixed-scope and fixed-price, quoted after a short Week 0 discovery, so you know the number before any work begins. Separately, the certification body's own Stage 1 and Stage 2 audit typically runs $10,000–$22,000 AUD for an SMB, with annual surveillance audits from around $2,500 in years one and two and a re-certification audit (roughly $9,000–$17,000) in year three. These external ranges are illustrative and depend on scope and the body you choose.

How is this different from an ISO 42001 course?

A course certifies a person to implement or audit the standard. This service certifies your company's AI Management System. Aegentra practitioners build the agreed AIMS scope inside your tenant and prepare you; any internal audit is a separate engagement with independence and conflict checks before acceptance for the Stage 1 and Stage 2 audit by an accredited certification body. If you want to train your own people, that runs separately through Aegentra Academy.

Does ISO 42001 make me EU AI Act compliant?

Not automatically. The EU AI Act is binding law; ISO 42001 is a voluntary international standard. Certification does not equal legal compliance on its own, but ISO 42001 is the most defensible AI-governance evidence base a provider or deployer can point to, and the CEN-CENELEC standards being written to support the Act draw heavily on it. For Australian companies selling AI into the EU, an ISO 42001 AIMS is the practical fast lane to demonstrating governance maturity.

Which AI systems go in scope?

You define the scope. It typically covers the AI use cases inside your product, the data pipelines feeding them, the model life cycle from development to retirement, the human-oversight controls, and the third-party AI you embed. In a Microsoft-based environment that usually means Microsoft 365 Copilot, Azure OpenAI, embedded model APIs, and any classifiers, recommendation engines, or generative features in your own product.

Do we have to be on Microsoft 365?

Our fastest path is for Microsoft 365 and Azure tenants, where most Australian SMB AI use already runs and where the Aeges risk engine collects evidence automatically. Other environments can be scoped during the Week 0 discovery — the AIMS framework is platform-agnostic; the speed advantage comes from building the controls where your AI actually lives.

Who issues the ISO 42001 certificate?

A JAS-ANZ-accredited certification body issues the certificate — Aegentra implements and prepares the AIMS but never certifies its own work, which preserves the independence auditors require. We sit in on both the Stage 1 and Stage 2 audits with you, and an optional retainer keeps the AIMS current as new models and use cases are deployed.

A worked example

The ISO 42001 implementation case study sets out eight AI governance mistakes we see repeatedly — a board-approved AI policy with no committee behind it, a human-in-the-loop control nobody measured, and an AI incident category that did not exist — along with the order the work was done in and why the inventory comes before the policy. It is an illustrative composite scenario written to demonstrate method, not a client engagement, and its figures describe no identifiable organisation. Audited engagement records, where every figure is the audited total from the engagement, are published in the Aegentra case studies.

Who will deliver your ISO 42001 implementation?

The AIMS scope identifies the accountable implementation lead and any specialist contribution needed for the AI systems, risks and governance processes inside scope.

• Accountable delivery — ISO 42001 implementation lead. The named lead, work packages, responsibilities and exclusions are recorded in the written scope.
• Specialist contribution — Assigned where the AI estate or agreed controls require it. The function and responsibility are disclosed before that specialist work begins.
• Assurance boundary — Separately scoped internal auditor and certification body. Aegentra does not present its own implementation work as an independent internal audit. The certification body decides certification.

Qualified delivery matched to your scope

Our team combines implementation, audit, security-governance and technical expertise. Across their professional careers, team members have delivered 2,000+ hours of ISO, ISM and GRC audit work across 70+ organisations.

The people assigned to your engagement—and the experience and qualification evidence relevant to their roles—are confirmed in your proposal before work begins.

Not every AIMS requires the same delivery roles. The assignment follows the agreed AI inventory, risk profile and implementation boundary. View our delivery-team capability.

Where we work

Delivered across Australia — all states and territories — with remote delivery and onsite attendance by arrangement. New Zealand and wider Asia Pacific engagements may be available by arrangement.

For organisations using Microsoft 365 Copilot, a Microsoft 365 permissions and data-protection review can support the technical work alongside the wider AIMS.