FAQGlobalISO/IEC 42001

ISO/IEC 42001 FAQ Generative AI

ISO/IEC 42001 can govern generative AI development, provision, procurement, and use through the same context-, risk-, impact-, control-, and lifecycle-based AIMS processes used for other AI systems.

The standard does not create a separate universal generative-AI control set. Select and supplement controls for the system's actual purpose, affected parties, data, resources, suppliers, outputs, misuse, and deployment context.

Author
Sorena AI
Published
May 9, 2026
Updated
Jul 24, 2026
Questions
4

Structured answer sets in this page tree.

Primary sources
2

Cited legal and guidance references.

Publication metadata
Sorena AI
Published May 9, 2026
Updated Jul 24, 2026
Overview

Register generative AI services and features inside the where they fall within scope. Generative AI produces new content such as text, images, audio, video, or software from user or system inputs; the label does not determine the risk level or the controls. Distinguish internally developed systems, model or API suppliers, embedded features, and employee tools because access to evidence and allocation of lifecycle responsibility differ.

Search this module

Find a question or answer quickly

4 of 4 questions
Question 1

How should teams handle Generative AI under ISO/IEC 42001?

Treat each generative AI use as a system in context, not as a model name or vendor account. Document the intended purpose, users and affected parties, prohibited or , decisions influenced by outputs, deployment environment, model and service versions, system prompts, retrieval sources, connected tools, input and output data, and supplier-imposed limits.

Classify the relationship before selecting controls. An organisation that trains or fine-tunes a model controls different lifecycle evidence from one that calls an external API, adds retrieval and tools to a supplied model, enables a feature inside business software, or permits staff to use a public service. For each case, record which controls the organisation operates, which evidence the supplier provides, which evidence is unavailable, and whether the remaining limitation is accepted, mitigated, or blocks the use.

Use the and risk assessment to select evaluation and oversight. Examples can include testing grounded answers against approved sources for a retrieval assistant, checking whether a coding assistant introduces insecure dependencies, testing demographic and language performance for a customer service tool, and preventing a content generator from exposing confidential prompts or producing prohibited impersonation. These are examples; the intended use and affected parties determine the actual tests.

Generative AI is not a separate ISO/IEC 42001 certification class, and it is not automatically a general-purpose AI model under the EU AI Act. Where EU law applies, assess the model, downstream system, operator role, and Chapter V or other obligations separately. ISO/IEC 42001:2023 is voluntary unless a contract, policy, or law requires its use. The standard has no universal statutory implementation deadline.

  • Record intended use, users, affected parties, decision consequences, and for each use case rather than approving a model for every purpose.
  • Map model, data, prompts, retrieval, tooling, hosting, access, logging, moderation, and supplier dependencies, including which party can change each component.
  • Set evaluation criteria for the actual use. Include task success, unsupported claims, harmful or prohibited content, privacy and security failures, and performance for relevant groups where those measures fit the impact analysis.
  • Define when a human must review, reject, override, or escalate an output, and name the authority that can restrict, suspend, or retire the use.
Citations
ISO/IEC 42001:2023 standard page

ISO/IEC 42001:2023 applies to organisations developing, providing, or using AI systems. Clauses 4, 6, and 8 and Annex controls A.4-A.10 ground scope, resources, lifecycle, data, information, responsible use, and supplier treatment for generative AI.

Regulation (EU) 2024/1689 (AI Act)

Articles 2 and 3 and Chapter V provide the separate binding EU scope, definitions, and general-purpose AI model obligations; the legal category does not follow from the marketing label 'generative AI'.

Question 2

What evidence should prove Generative AI is current under ISO/IEC 42001?

Evidence should match the selected controls and the released use case. Keep an inventory record; owner and approved users; model, service, prompt, retrieval, data, tool, and hosting versions; supplier documentation and accepted limitations; risk and impact assessments; test sets and acceptance criteria; release decision; user instructions; human-review samples; event logs; monitoring results; complaints and incidents; provider change notices; and reassessment decisions.

Preserve enough detail to reproduce the conclusion. Record the evaluation date, system configuration, sample source and limits, relevant user or affected group, metric and threshold, result, reviewer, exceptions, and approval. A vendor benchmark or model card may support the record but does not prove that the configured application works for the organisation's data, users, tools, and decisions.

  • Record what supplier evidence was available, what was unavailable, and how each accepted limitation affects use.
  • Sample real outputs, human review, overrides, failures, complaints, and escalations against documented acceptance criteria without retaining personal or confidential content longer than authorised.
  • Test connected tools and retrieval separately from the base model because access rights, source freshness, tool execution, and prompt handling can change the system's risk.
  • Link model, terms, data, retrieval, prompt, tool, moderation, or service changes to a documented decision to retest, reassess, restrict, or continue the use.
Citations
ISO/IEC 42001:2023 standard page

ISO/IEC 42001:2023 Clauses 7.5 and 8 require controlled documented information and operating evidence; Annex A.6-A.10 addresses lifecycle records, data, user information, responsible use, and suppliers.

Recommended next step

Put the generative AI guidance into practice

Capture owners, evidence, decisions, and review dates in one workflow record so AI governance controls and escalation points stay auditable over time.

Question 3

Who should approve Generative AI decisions under ISO/IEC 42001?

Assign a business owner for the use and its outcomes, technical owners for integration and monitoring, data and security owners where relevant, procurement for supplier commitments, and an authority able to restrict, suspend, or retire the use when controls fail.

The designated management authority approves the risk-treatment plan and residual AI risk. Legal, privacy, security, safety, employment, intellectual-property, records, or sector owners should decide issues within their own authority; approval cannot waive a binding requirement or replace those decisions.

For supplier services, assign who reviews terms and documentation, approves data sent to the service, receives model or policy changes, controls accounts and integrations, monitors the configured use, reports incidents, and executes exit or data-return arrangements.

  • Allocate internal and supplier responsibilities for data, models, integration, user information, monitoring, incidents, updates, and support.
  • Separate technical evaluation and procurement from residual-risk acceptance and final use approval.
  • Keep approval records with the evidence rather than in disconnected email threads.
Citations
ISO/IEC 42001:2023 standard page

ISO/IEC 42001:2023 Clauses 5.3 and 6.1.3 require assigned authorities and management approval of treatment and residual risk; Annex A.10 requires lifecycle responsibility allocation across third parties.

Question 4

When should Generative AI be reviewed under ISO/IEC 42001?

Review at planned intervals and when the model or service version, terms, data sources, system prompt, retrieval design, tools, moderation, intended use, user population, deployment context, output reliance, supplier evidence, incidents, or applicable requirements change.

A supplier's model update can change behaviour without an application-code release. Route provider notices, silent evaluation drift, new limitations, changed usage, and unexpected tool behaviour through the same risk, impact, approval, and monitoring controls as an internal significant change.

Reassess before a pilot becomes production, a low-consequence drafting aid begins influencing decisions about people, an internal tool becomes customer-facing, new data categories are entered, or a component gains access to external actions. Restrict or retire the use when required evidence is unavailable or residual risk no longer meets the approved criteria.

  • Use planned reviews plus event-driven triggers.
  • Reassess impacts and controls after significant changes.
  • Restrict or retire uses whose residual risk is no longer accepted.
Citations
ISO/IEC 42001:2023 standard page

ISO/IEC 42001:2023 Clauses 8.2-8.4 require reassessment at planned intervals and around significant changes; Annex B.6.2.6 explains that performance can change because of production-data drift even without continuous learning.

Primary sources

References and citations

iso.org
Referenced sections
  • ISO/IEC 42001:2023 Clauses 8.2-8.4 require reassessment at planned intervals and around significant changes; Annex B.6.2.6 explains that performance can change because of production-data drift even without continuous learning.
"requirements for establishing, implementing, maintaining and continually improving an Artificial Intelligence Management System"
eur-lex.europa.eu
Referenced sections
  • Articles 2 and 3 and Chapter V provide the separate binding EU scope, definitions, and general-purpose AI model obligations; the legal category does not follow from the marketing label 'generative AI'.
Related guides

Explore more topics

ISO/IEC 42001 AI Impact Assessment Template
ISO/IEC 42001 AI-system impact assessment template for consequences, affected people, foreseeable misuse, context, evidence, approval, and reassessment.
ISO/IEC 42001 AI Management FAQ
Plain-language ISO/IEC 42001 FAQ covering scope, policy, Annex controls, risk and impact assessment, suppliers, monitoring, certification, and legal limits.
ISO/IEC 42001 AI Policy FAQ
ISO/IEC 42001 AI policy requirements, approval, communication, evidence, alignment with other policies, and review triggers.
ISO/IEC 42001 AI System Inventory Guide
Build an ISO/IEC 42001 AI inventory covering purpose, owners, lifecycle roles, data, resources, suppliers, impacts, risks, controls, and monitoring.
ISO/IEC 42001 AI System Inventory Workflow
ISO/IEC 42001 workflow for creating, approving, maintaining, changing, and retiring AI-system inventory records with accountable evidence.
ISO/IEC 42001 AIMS Scope Decision Guide
Define an auditable ISO/IEC 42001 AIMS boundary across organisational units, AI activities, products, services, interfaces, suppliers, and customers.
ISO/IEC 42001 AIMS Scope Decision Workflow
ISO/IEC 42001 AIMS scope workflow for context, interested parties, AI activities, external dependencies, boundary approval, and change review.
ISO/IEC 42001 Certification FAQ
ISO/IEC 42001 certification scope, readiness evidence, internal audit, management review, corrective action, and claim limits.
ISO/IEC 42001 Compliance Guide
ISO/IEC 42001 conformance guide for Clauses 4-10, risk treatment, controls, operating evidence, internal audit, management review, and correction.
ISO/IEC 42001 Controls and Governance Model Guide
ISO/IEC 42001 governance model linking leadership, policy, risk and impact assessment, Annex controls, lifecycle roles, monitoring, audit, and improvement.
ISO/IEC 42001 High Risk AI FAQ
Separate ISO/IEC 42001 organisational risk criteria from legal high-risk AI classifications, with evidence, approval, and reassessment triggers.
ISO/IEC 42001 Human Oversight FAQ
Design and evidence effective human oversight under ISO/IEC 42001, including competence, information, intervention authority, testing, and review.
ISO/IEC 42001 Model Monitoring Evidence Guide
ISO/IEC 42001 monitoring evidence for AIMS performance, AI-system outcomes, impacts, control effectiveness, supplier signals, escalation, and correction.
ISO/IEC 42001 Post Market Monitoring FAQ
Distinguish ISO/IEC 42001 operational monitoring from legal post-market monitoring and connect real-world evidence to risk, review, incidents, and correction.
ISO/IEC 42001 Provider and Deployer Roles FAQ
Map ISO/IEC 42001 lifecycle responsibilities across developers, users, suppliers, customers, and third parties while keeping legal operator roles separate.
ISO/IEC 42001 Requirements Guide
ISO/IEC 42001:2023 requirements explained across Clauses 4-10, Annex A controls, Annex B guidance, evidence, audit, review, and improvement.
ISO/IEC 42001 Risk Controls FAQ
Select, justify, approve, operate, and review ISO/IEC 42001 risk controls, including Annex A comparison, the statement of applicability, residual risk, and evidence.
ISO/IEC 42001 vs EU AI Act Comparison
Compare voluntary ISO/IEC 42001 AIMS certification with binding EU AI Act roles, classifications, duties, evidence, dates, and enforcement.
ISO/IEC 42001 vs ISO/IEC 23894 Comparison
Compare certifiable ISO/IEC 42001 AIMS requirements with ISO/IEC 23894 AI risk-management guidance and see how to integrate their evidence.
ISO/IEC 42001 vs NIST AI RMF Comparison
Compare ISO/IEC 42001 AIMS requirements with the voluntary NIST AI RMF GOVERN, MAP, MEASURE, and MANAGE functions and evidence.