Start hereInventory AI and define governance scope
~119 days leftOpen step
Guide library
Certification Global

ISO 42001

ISO/IEC 42001:2023 · AI management systems

Put AI governance, impact assessment and lifecycle controls into a certifiable management system.

Who it applies to

Organizations developing, providing or using AI that want an AI management system in a defined scope.

What you’ll produce

AI management system and optional third-party certification for that management system scope.

Map legal obligations independently. A management system certificate has a defined organizational scope.

TEAM KICKOFF

Run the project

Go to the roadmap

Decisions for leadership

What do we need the certification to cover?
Confirm the applicable edition, sites, products or service boundary and the certification route. Budget for assessment, changes needed to pass, and maintaining certification.
Who will do the work and who can sign?
Appoint a sponsor, a day-to-day lead and a named owner for every workstream. Set aside their time and decide who can approve the outcome, accept remaining risk or authorize a filing.
Which deadlines can we promise?
Separate fixed external dates from internal targets. Confirm dependencies such as a vendor, test lab or assessor, and approve a backup plan if their dates slip.

How to spend the first week

  1. Ask the first-step owner to bring the most recent system, activity or product inventory and the contract or rule driving the work.
  2. Ask the appropriate assessor or program contact what scope they will review, what evidence they need, and which dates and fees they can confirm in writing.
  3. Record what the sponsor approved, open questions, work owners and the next date a decision is needed. Send the brief to everyone who needs to contribute evidence.

What to review each week

Monitor unresolved scope questions, controls without evidence, findings that need a retest and the next assessment or renewal date.

Kickoff brief to share with your team

Fill this in with your sponsor and send it to the people who own the work.

ISO 42001 kickoff brief
Business reason and required outcome  [Contract, obligation or approval]
Scope and exclusions  [Entities, systems or activities, and reasons]
Fixed deadlines and internal milestones  [Date, source and owner]
Executive sponsor and authorized signer  [Names]
Project lead and backup  [Names]
First workstream  Inventory AI and define governance scope · AI governance lead + leadership [Name and due date]
People, effort and external costs  [Estimate, owner and contingency]
Evidence folder and reviewer access  [Location, access owner and review date]
Decisions and blockers  [Item, options, owner and decision date]

First week
1. Ask the first-step owner to bring the most recent system, activity or product inventory and the contract or rule driving the work.
2. Ask the appropriate assessor or program contact what scope they will review, what evidence they need, and which dates and fees they can confirm in writing.
3. Record what the sponsor approved, open questions, work owners and the next date a decision is needed. Send the brief to everyone who needs to contribute evidence.
How to hand off evidence
  1. Record the requirement or claim, the person responsible, the period and the result. Link to the original report, configuration, test or case record.
  2. Include the version, system or product, date, and how you gathered the record. When a report covers a group of items, retain its filters and reconcile the count to the source.
  3. Ask the reviewer to open the evidence and reproduce the key check. Record their decision and date. Keep missing items or failures visible with an owner, a due date and a retest when needed.
  4. Share through the agreed channel. Keep sensitive originals in a restricted location and link any redacted copies to the corresponding originals.
The whole journeyabout 119 days across 4 steps
  1. 01Inventory AI and define governance scopeAI governance lead + leadership · ~14 days
  2. 02Assess risks and impactsAI risk lead + affected teams · ~35 days
  3. 03Evidence lifecycle controlsProduct + model owners · ~42 days
  4. 04Audit and improve the management systemInternal audit + leadership · ~28 days
IMPLEMENTATION ROADMAP

Step by step

Checklist, instructions and 12 templates.

View plan as a table
Project roadmap with current target dates and completion status
StepOwnerWhen to actEstimated targetDeliverablesStatus
Inventory AI and define governance scopeAI governance lead + leadershipAt initiationSet a start dateAI inventory; Scope statement; AI policyOpen
Assess risks and impactsAI risk lead + affected teamsBefore deployment and after material changeSet a start dateAI risk register; Impact assessments; Control applicabilityOpen
Evidence lifecycle controlsProduct + model ownersAcross the lifecycleSet a start dateValidation reports; Monitoring plan; Supplier controlsOpen
Audit and improve the management systemInternal audit + leadershipBefore external audit and on recurring cadenceSet a start dateInternal audit; Management review; Corrective actionsOpen
0 of 4 completeOpen workspace

Targets use illustrative calendar-day estimates and assume the steps run in sequence. Check official dates and agree real durations with your team.

  1. 01

    Inventory AI and define governance scope

    Who AI governance lead + leadership When At initiation

    Identify systems, lifecycle roles, dependencies and intended contexts. Approve AI policy and allocate decision rights.

    What to produce

    AI inventorySpreadsheet

    Gather first

    • Asset exports or system discovery
    • Service owner interviews
    • Approved scope and architecture

    How to prepare it

    1. Export a dated baseline from the source system. Give each asset, service or record class a stable ID and remove duplicates without losing links to the source.
    2. Add ownership, purpose, sensitivity and any vendor or system it depends on. Mark unknowns explicitly and assign them to someone to resolve.
    3. Reconcile the register against an independent source such as cloud billing or the production deployment list. Set a refresh cadence and an owner.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    ID
    SYS-014
    Item
    Production billing database
    Owner
    Platform operations
    Location and purpose
    Production EU region, stores invoice and account data
    Impact and dependencies
    High. Identity provider and managed backups.
    Verified on
    2026-11-10, cloud inventory export

    Check before handing it over

    Each item has one accountable owner. The source and as-of date let a reviewer trace it back.

    Share the reviewed inventory with the people building the control, privacy or testing scope.

    Scope statementMemo

    Gather first

    • Service inventory and architecture
    • Customer commitments and contracts
    • Applicable criteria or legal text

    How to prepare it

    1. Write the exact decision this memo supports. Name the legal entity, service, environments and dates. Attach a boundary diagram and state which people, vendors and processes can affect the scope.
    2. List the exclusions next to the inclusion list. Give a reason for each one and describe any shared service that crosses the boundary.
    3. Review the draft with service owners and the person making the decision. Freeze the approved version and add a review date or trigger for changes.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Decision
    Cover the hosted billing service for the April to June review.
    In scope
    Production billing API, hosting, releases and production access.
    Excluded and why
    Internal HR tool. It has no access to billing systems or data.
    Dependencies
    Cloud hosting and corporate identity provider.
    Approval
    Service owner approves after checking the architecture and contract.

    Check before handing it over

    A reviewer can use the memo to classify a named system and see how shared dependencies are handled.

    Give the approved scope and any open questions to whoever will map requirements, select samples or review the engagement.

    AI policyMemo

    Gather first

    • Approved scope
    • Governing requirements and risks
    • Teams, service dependencies and review gates

    How to prepare it

    1. State the objective, scope, accountable owner and conditions that trigger action. Translate commitments into tasks with periods, thresholds and evidence.
    2. For each task set the input, decision rule, output and handoff. Include escalation for failures and approvals for exceptions.
    3. Walk through one real case with owners and reviewers. Approve the version, schedule review, and check that promised tasks have operational records.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Purpose
    Identify failed production backup jobs and restore service data.
    Scope
    Production billing databases and backup vault
    Activity and timing
    On-call reviews daily job alerts and runs a quarterly restore test.
    Evidence and exception
    Keep job records and restore test output. Page the service owner if a job fails.
    Owner and approval
    Operations lead owns. Service owner approves.

    Check before handing it over

    A person who receives the plan can tell when to act, what to check and where to save the result.

    Give the approved plan to the people responsible for operation. Add dated checkpoints to the working calendar.

  2. 02

    Assess risks and impacts

    Who AI risk lead + affected teams When Before deployment and after material change

    Evaluate impacts on individuals, groups and society and select relevant controls. Document risk treatment and review thresholds.

    What to produce

    AI risk registerSpreadsheet

    Gather first

    • Processing or service design and affected people
    • Threats, consequences and assessment triggers
    • Risk methodology and current safeguards

    How to prepare it

    1. Screen the activity against the applicable assessment trigger. Describe the real scenario and who or what could be harmed. Record likelihood and consequence using your criteria.
    2. Describe existing safeguards. Decide what reduces the risk and what risk remains after those measures. For a required impact assessment, document necessity, alternatives and consultation where applicable.
    3. Send the result to the person who can approve, change or stop the activity. Set a review trigger and keep evidence of the decision.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Activity and risk
    Account recovery could expose invoices to an impersonator.
    Affected people or assets
    Customer account contacts and invoices
    Current controls
    Recovery uses verified channel. Sampled evidence linked.
    Proposed treatment
    Require a second check for requests to change the billing address.
    Remaining risk and decision
    Moderate under company method. Approval required before launch.
    Owner and review trigger
    Product risk owner, repeat when recovery flow changes

    Check before handing it over

    The assessment explains the potential harm, why each measure helps and who accepts the remaining risk.

    Give the signed review to the owner of the launch or processing decision. Use the regulator’s assessment form where one applies.

    Impact assessmentsSpreadsheet

    Gather first

    • Processing or service design and affected people
    • Threats, consequences and assessment triggers
    • Risk methodology and current safeguards

    How to prepare it

    1. Screen the activity against the applicable assessment trigger. Describe the real scenario and who or what could be harmed. Record likelihood and consequence using your criteria.
    2. Describe existing safeguards. Decide what reduces the risk and what risk remains after those measures. For a required impact assessment, document necessity, alternatives and consultation where applicable.
    3. Send the result to the person who can approve, change or stop the activity. Set a review trigger and keep evidence of the decision.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Activity and risk
    Account recovery could expose invoices to an impersonator.
    Affected people or assets
    Customer account contacts and invoices
    Current controls
    Recovery uses verified channel. Sampled evidence linked.
    Proposed treatment
    Require a second check for requests to change the billing address.
    Remaining risk and decision
    Moderate under company method. Approval required before launch.
    Owner and review trigger
    Product risk owner, repeat when recovery flow changes

    Check before handing it over

    The assessment explains the potential harm, why each measure helps and who accepts the remaining risk.

    Give the signed review to the owner of the launch or processing decision. Use the regulator’s assessment form where one applies.

    Control applicabilitySpreadsheet

    Gather first

    • Applicable criteria, requirements or risk decisions
    • Current procedures and configurations
    • Names of people operating or reviewing each control

    How to prepare it

    1. Give each requirement and implemented control an ID. Write a short control statement naming who does what, when and in which system.
    2. Record the evidence the control produces and how it could be tested. For a shared or inherited control, state who owns each part.
    3. Walk through one control with its operator. Check a real recent run, record gaps and obtain approval for exclusions or alternate controls.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Requirement
    Production access is reviewed periodically.
    Control ID
    AC-04
    Control activity
    IAM team exports administrator roles every quarter. System owners approve or revoke each assignment.
    Owner and frequency
    IAM lead, quarterly
    Evidence
    Export with filters, annotated review and removal tickets
    Test and status
    Reconcile export to current roles. Inspect approvals. Operating.

    Check before handing it over

    A different person can operate the control or test it using the matrix and its evidence links.

    Give the baseline to control owners and the assessor before walkthroughs or sampling.

  3. 03

    Evidence lifecycle controls

    Who Product + model owners When Across the lifecycle

    Define acceptance criteria, data governance, evaluation, oversight, monitoring, incident response and supplier obligations. Save actual operation records.

    What to produce

    Validation reportsMemo

    Gather first

    • Approved test scope and criteria
    • Population and sample records
    • Test approach, results and reviewer requirements

    How to prepare it

    1. State objective, dates, scope and the method before describing results. Record the tester, any independence requirements, and how samples were selected.
    2. For each item, compare what should have happened with what the evidence shows. Describe impact and link the evidence or recorded exception.
    3. Add limitations, obtain review and give each finding an owner and a date. Retain the underlying samples and calculations with the signed version.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Test objective
    Check that production role removals happen within the internal deadline.
    Method and population
    All September departures. Test each privileged account.
    Expected result
    Remove access within one business day.
    Observed result
    11 on time. One late by two days, ticket IAM-114.
    Finding and follow-up
    F-008. Retest the next departure and verify alerting.

    Check before handing it over

    Someone who did not run the test can reproduce the result and tell a finding from a scope limitation.

    Give management the findings and evidence links. Share observations with an external reviewer where the engagement calls for it.

    Monitoring planMemo

    Gather first

    • Approved scope
    • Governing requirements and risks
    • Teams, service dependencies and review gates

    How to prepare it

    1. State the objective, scope, accountable owner and conditions that trigger action. Translate commitments into tasks with periods, thresholds and evidence.
    2. For each task set the input, decision rule, output and handoff. Include escalation for failures and approvals for exceptions.
    3. Walk through one real case with owners and reviewers. Approve the version, schedule review, and check that promised tasks have operational records.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Purpose
    Identify failed production backup jobs and restore service data.
    Scope
    Production billing databases and backup vault
    Activity and timing
    On-call reviews daily job alerts and runs a quarterly restore test.
    Evidence and exception
    Keep job records and restore test output. Page the service owner if a job fails.
    Owner and approval
    Operations lead owns. Service owner approves.

    Check before handing it over

    A person who receives the plan can tell when to act, what to check and where to save the result.

    Give the approved plan to the people responsible for operation. Add dated checkpoints to the working calendar.

    Supplier controlsSpreadsheet

    Gather first

    • Applicable criteria, requirements or risk decisions
    • Current procedures and configurations
    • Names of people operating or reviewing each control

    How to prepare it

    1. Give each requirement and implemented control an ID. Write a short control statement naming who does what, when and in which system.
    2. Record the evidence the control produces and how it could be tested. For a shared or inherited control, state who owns each part.
    3. Walk through one control with its operator. Check a real recent run, record gaps and obtain approval for exclusions or alternate controls.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Requirement
    Production access is reviewed periodically.
    Control ID
    AC-04
    Control activity
    IAM team exports administrator roles every quarter. System owners approve or revoke each assignment.
    Owner and frequency
    IAM lead, quarterly
    Evidence
    Export with filters, annotated review and removal tickets
    Test and status
    Reconcile export to current roles. Inspect approvals. Operating.

    Check before handing it over

    A different person can operate the control or test it using the matrix and its evidence links.

    Give the baseline to control owners and the assessor before walkthroughs or sampling.

  4. 04

    Audit and improve the management system

    Who Internal audit + leadership When Before external audit and on recurring cadence

    Conduct internal audit, complete management review and address findings. If needed, engage a qualified certification body.

    What to produce

    Internal auditMemo

    Gather first

    • Approved test scope and criteria
    • Population and sample records
    • Test approach, results and reviewer requirements

    How to prepare it

    1. State objective, dates, scope and the method before describing results. Record the tester, any independence requirements, and how samples were selected.
    2. For each item, compare what should have happened with what the evidence shows. Describe impact and link the evidence or recorded exception.
    3. Add limitations, obtain review and give each finding an owner and a date. Retain the underlying samples and calculations with the signed version.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Test objective
    Check that production role removals happen within the internal deadline.
    Method and population
    All September departures. Test each privileged account.
    Expected result
    Remove access within one business day.
    Observed result
    11 on time. One late by two days, ticket IAM-114.
    Finding and follow-up
    F-008. Retest the next departure and verify alerting.

    Check before handing it over

    Someone who did not run the test can reproduce the result and tell a finding from a scope limitation.

    Give management the findings and evidence links. Share observations with an external reviewer where the engagement calls for it.

    Management reviewMemo

    Gather first

    • Agenda or decision request
    • Current performance measures and findings
    • Proposed actions and participants

    How to prepare it

    1. Write what decision is needed and list the reports or records the decision maker reviewed. Capture the date, people and decision authority.
    2. Record the decision, alternatives considered, any remaining risk and exactly who will do what by when.
    3. Send notes for factual review. File the approved record and add actions to the working tracker so progress reaches the next review.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Item
    Approve a fix for the missing access-removal retry.
    Inputs reviewed
    September test results and IAM incident root-cause notes.
    Decision
    Approve automatic retry plus daily reconciliation.
    Action and due date
    Platform lead to ship and test by 2026-11-28.
    Residual risk and follow-up
    Manual review remains until rollout. Check at December meeting.

    Check before handing it over

    The minutes identify the person who approved each action and the date the decision will be revisited.

    Give owners the actions and add the approved record to the relevant audit or governance file.

    Corrective actionsSpreadsheet

    Gather first

    • Finding, risk or test exception
    • Root-cause analysis
    • Named owner, due date and expected evidence

    How to prepare it

    1. Record the observation and link the exact test or requirement. Describe the business effect and investigate why the issue was able to happen.
    2. Separate immediate containment from the permanent fix. Give each action an owner, date, verification method and any required approval for deferral.
    3. Retest against the original failure once the fix is live. Record who accepted the outcome and keep unresolved issues visible.

    Worked example

    Fictional details. Use the same fields with evidence from your work.

    Finding or risk
    F-008, a departed contractor retained production access
    Cause and effect
    HR event never reached IAM. Admin access remained active.
    Action and owner
    Revoke access and alert on failed HR sync. IAM lead.
    Due and status
    2026-11-28, in progress
    Closure test
    Simulate departure and verify access disappears within policy
    Approver and evidence
    Risk owner, ticket IAM-114 and retest results

    Check before handing it over

    The closure test addresses the original problem and has a dated result from an independent reviewer where required.

    Send open items and decisions to governance, audit or the assessor on the applicable cadence.

PRIMARY REFERENCES

Official sources

How we build these guides

Keyboard shortcuts

⌘K or /
Search everything
?
This list
g h
Home
g l
Guide library
g f
Find my path
g t
Templates
g p
Playbooks
g w
My workspace
g d
Deadline calendar
g c
Regulatory changes