Comparison · International standards
ISO 27001 vs ISO 42001
ISO/IEC 27001:2022 (information security) and ISO/IEC 42001:2023 (AI management) side by side, on the same criteria, with the clause or Annex A reference behind every cell. Written for a company that holds 27001 and is being asked about 42001 by a customer, or is deciding whether to extend one management system rather than build two.
Sources: ISO/IEC 27001:2022, Information security management systems: Requirements; ISO/IEC 42001:2023, Artificial intelligence management system: Requirements; ISO/IEC 42006:2025, Requirements for bodies providing audit and certification of AI management systems; Digital Personal Data Protection Rules, 2025, G.S.R. 846(E)
Side by side
Thirteen criteria, reviewed 4 Nov 2026. Each cell names its source. Scroll sideways on a phone.
| Criterion | ISO/IEC 27001:2022 | ISO/IEC 42001:2023 |
|---|---|---|
| Subject | Information security: preserving the confidentiality, integrity and availability of information through an information security management system (ISMS).clause 1; ISO/IEC 27000 | Responsible development, provision and use of AI systems through an AI management system (AIMS). Applies to any organisation that develops, provides or uses AI, in any sector.clause 1 |
| Edition and dates | Third edition, published 25 Oct 2022, with Amd 1:2024 adding climate change wording. The transition from the 2013 edition ended 31 Oct 2025.ISO/IEC 27001:2022; Amd 1:2024 | First edition, published Dec 2023. Requirements for the bodies that audit and certify against it followed in ISO/IEC 42006:2025.ISO/IEC 42001:2023; ISO/IEC 42006:2025 |
| Structure | ISO's harmonised structure for management system standards: clause 4 context, 5 leadership, 6 planning, 7 support, 8 operation, 9 performance evaluation, 10 improvement.clauses 4 to 10 | The same harmonised structure, clauses 4 to 10, so one management system can carry both. Each standard still needs its own internal audit and management review.clauses 4 to 10, 9.2, 9.3; Annex D.2 |
| Controls | 93 reference controls in Annex A, in four themes: organisational (37), people (8), physical (14), technological (34).Annex A | 38 reference controls in Annex A, in nine groups from A.2 (policies related to AI) to A.10 (third-party and customer relationships). Annex B gives implementation guidance for each.Annex A; Annex B |
| Statement of Applicability | Required. Lists the necessary controls, whether they are implemented, and the justification for including or excluding each Annex A control.clause 6.1.3 | Required, by the same mechanism. Annex B guidance must be taken into account when implementing controls, but the organisation need not justify adopting or departing from it in the statement.clause 6.1.3; Annex B |
| Risk assessment | Information security risk assessment: define criteria, identify risks to confidentiality, integrity and availability, analyse and evaluate them, then treat them under a plan with risk owners.clauses 6.1.2, 6.1.3, 8.2, 8.3 | AI risk assessment against AI-specific objectives and risk sources. Annex C lists example objectives (fairness, safety, security, privacy, transparency and explainability, robustness and others) and risk sources; the organisation decides which apply.clauses 6.1.2, 6.1.3; Annex C |
| Impact assessment | None as a separate requirement. Impact on the organisation is one input to analysing a risk.clause 6.1.2 | A separate AI system impact assessment: assess the consequences for individuals, groups and society of developing, providing or using an AI system, document the results, and keep them.clause 6.1.4; Annex A 5.2 to 5.5 |
| Lifecycle controls | Secure development lifecycle, application security requirements, secure architecture and coding, security testing, outsourced development, separation of environments, change management.Annex A 8.25 to 8.32 | AI system lifecycle: objectives for responsible development, design and development process, requirements and specification, design documentation, verification and validation, deployment, operation and monitoring, technical documentation, event logging.Annex A 6.1.2 to 6.2.8 |
| Data | Data as an information asset: classification, labelling, transfer, deletion, masking, leakage prevention and backup.Annex A 5.12 to 5.14, 8.10 to 8.13 | Data for AI systems: a data management process, acquisition, quality, provenance and preparation of the data used to train, validate and test models.Annex A 7.2 to 7.6 |
| Telling people outside | Communication with interested parties is planned under clause 7.4; there is no duty to explain how a system works to its users.clause 7.4 | Information for interested parties: system documentation and information for users, a channel for external reporting, communication of incidents, and information for other interested parties.Annex A 8.2 to 8.5 |
| Suppliers and customers | Supplier relationships, security terms in supplier agreements, the ICT supply chain, monitoring of supplier services and use of cloud services.Annex A 5.19 to 5.23 | Third-party and customer relationships: allocate responsibilities across the AI supply chain, manage suppliers of AI components and data, and meet customer expectations for the AI system.Annex A 10.2 to 10.4 |
| Certification | By a certification body accredited under ISO/IEC 17021-1, with ISO/IEC 27006-1 setting the sector-specific requirements for ISMS auditors and audit time.ISO/IEC 17021-1; ISO/IEC 27006-1 | By a certification body accredited under ISO/IEC 17021-1, with ISO/IEC 42006:2025 setting the sector-specific requirements for AIMS auditors. Accredited 42001 certificates are newer and fewer; check the certification body's accreditation scope.ISO/IEC 17021-1; ISO/IEC 42006:2025 |
| Indian regulatory hooks | SEBI's Cybersecurity and Cyber Resilience Framework requires ISO 27001 certification for market infrastructure institutions and qualified regulated entities. DPDP Rule 6 names safeguards (encryption, access control, logs, backups, one-year log retention) that map onto Annex A.SEBI circular CIR/2024/113 (CSCRF), 20 Aug 2024; DPDP Rule 6 | No Indian law or regulator requires it. MeitY's India AI Governance Guidelines of Nov 2025 are voluntary and principle-based. Demand comes from customer contracts and from vendors selling AI into the EU.India AI Governance Guidelines, MeitY, Nov 2025 |
The differences that matter
Same skeleton, different risk
Both standards use the harmonised clauses 4 to 10, so scope, leadership, objectives, competence, internal audit and management review can be shared. What changes is the risk being managed: 27001 protects information from loss of confidentiality, integrity and availability; 42001 manages the harm an AI system can do to the people it affects, including bias, opacity and unsafe behaviour. A security team can run the first; the second needs product, data science and legal in the room.
The impact assessment is new work
27001 has no equivalent of clause 6.1.4. An AI system impact assessment looks outward at individuals, groups and society rather than inward at the organisation, and Annex A controls 5.2 to 5.5 turn it into a documented, repeated process. Companies that already do data protection impact assessments under GDPR Art. 35 or DPDP s. 10(2)(c) have a head start on the method but not on the questions.
Thirty-eight controls, mostly about process and people
27001's 93 controls are heavy on technical and physical measures. 42001's 38 are about policy, roles, resources, lifecycle discipline, data provenance, documentation and what you tell users and customers. Few of them are satisfied by configuration; most are satisfied by evidence that a decision was made, recorded and reviewed.
Who is asking, and why
27001 is asked for by Indian regulators and by procurement everywhere. 42001 is asked for by customers writing AI clauses into contracts and by vendors who need to show governance to EU buyers under the AI Act. No Indian regulator requires 42001 yet; MeitY's AI governance guidelines are voluntary. Treat it as a market signal, not a compliance obligation, and scope it to the AI systems you actually sell or run.
Questions
We are ISO 27001 certified. How much of ISO 42001 do we already have?
The management system shell: context, leadership, planning, support, operation, performance evaluation and improvement are the same clauses, and Annex D.2 of 42001 says an existing ISMS can carry the AIMS. What you add is an AI policy, AI-specific risk and impact assessments under clauses 6.1.2 to 6.1.4, the 38 Annex A controls that apply to you, and a separate internal audit and management review for the AIMS.
Can one certification body audit both in one visit?
Yes, if the body is accredited for both standards. ISO/IEC 42006:2025 sets the requirements for bodies certifying AI management systems, and accreditation bodies have been extending scopes since it was published. Integrated audits are common for 27001 with 27701 or 9001; ask the body for its accreditation scope before you plan one with 42001.
Does ISO 42001 make us compliant with the EU AI Act or Indian law?
No. It is a management system standard, not a product conformity standard, and certification to it is not a legal presumption of conformity with any law. It does produce much of the evidence a regulator or customer would ask for: risk and impact assessments, data provenance, documentation, logging and incident communication. India has no binding AI law; the DPDP Act applies to the personal data an AI system processes, and that is governed by DPDP, not by 42001.
Which should an Indian AI startup do first?
27001 first, almost always. Every enterprise buyer asks for it, SEBI requires it for regulated market entities, and it covers the data an AI product runs on. Add 42001 when a customer contract names it or when you sell a model or AI feature into the EU. Build the ISMS so the AIMS can sit on top of it rather than beside it.
Read next
This page explains the law in plain English. It is not legal advice.
Weekly · Watching 6 regulators