ISO 42001 verifiable records for your AI management system.
The standard requires controlled, retained records and recorded event logs (Clauses 7.5 and 9, control A.6.2.8). ar.io anchors them to permanent storage, so the evidence an auditor samples is provably the same record that was created at the time.
- StandardISO/IEC 42001AI Management System (AIMS)
- PublishedDec 2023First international AI MSS
- Annex A9 / 38Control objectives / controls
- CertificationVoluntaryVia accredited bodies
- Recert cycle3 yearsWith annual surveillance
Published by ISO and IEC. Certification is awarded by accredited certification bodies.
What ISO 42001 requires, in one block.
ISO/IEC 42001:2023 is the first international AI Management System (AIMS) standard. It specifies requirements (Clauses 4 to 10) for establishing, maintaining, and improving how an organization governs AI, supported by a normative Annex A of 9 control objectives and 38 controls. It applies to any organization that develops, provides, or uses AI systems. Certification is voluntary and awarded by accredited bodies, but it is increasingly required in enterprise procurement and aligns with EU AI Act governance expectations.
The clauses and controls that generate evidence.
Clauses 4 to 10 are the management system; Annex A is a normative catalogue you select from and justify in a Statement of Applicability. The requirements below are the ones that produce records an auditor samples, and the ones that pair with EU AI Act evidence obligations.
Documented information
Documented information the AIMS requires must be created, controlled, version-managed, access-controlled, and retained.
A controlled record set with provable version history and retention, so an auditor can trust the records have not changed since approval.
Monitoring & measurement
Monitor, measure, analyse, and evaluate the performance of AI systems and the AIMS itself.
Retained system event logs, performance-evaluation reports, and tracked KPIs kept as documented information.
Internal audit
Conduct internal audits at planned intervals and keep records of the results and any follow-up.
Audit plans, checklists, reports, findings, and corrective-action records, available at every surveillance audit.
Event logs
Record event logs across the AI system life cycle as part of the Annex A life-cycle controls.
Machine-generated event logs, retained and controlled so they remain trustworthy evidence over time.
Impact assessments
Assess the impacts of AI systems on individuals and society, and document the assessment.
Completed AI system impact assessments, retained as part of the evidence base.
Data provenance
Manage data for AI systems, including provenance and quality across the life cycle.
Data-provenance and data-quality records that establish where training and operating data came from.
Proof without access, in five steps.
Cryptographic fingerprints of the records your AIMS produces are generated inside your environment and anchored to permanent storage. The underlying documents never leave your perimeter. Only the fingerprint and timestamp are anchored. The architecture is called proof without access.
- 01
Capture
A REST client or MLflow plugin hashes the record (impact assessment, technical doc, event log, audit report) inside your environment.
- 02
Sign
The fingerprint is signed with your private key, authenticating the commitment.
- 03
Anchor
The signed fingerprint is written to Arweave through the ar.io gateway network. Records cannot be silently altered without leaving evidence.
- 04
Bundle
The evidence bundle (records, hashes, timestamps, chain of custody) exports in formats designed for certification and internal-audit review.
- 05
Verify
An auditor or customer compares fresh fingerprints to anchored ones using only your record and the public entry. No vendor cooperation required.
The clauses and control that turn on record integrity.
- Clause7.5Documented information kept controlled and provably unaltered.
- Clause9.1Monitoring evidence and event logs anchored as they are generated.
- Clause9.2Internal-audit records preserved and independently verifiable.
- ControlA.6.2.8Event logs anchored across the AI system life cycle.
The architecture meets the three properties auditors test for: existence at the time, tamper-evidence, and self-authentication. ar.io is not a certification body and not an AIMS; it is the integrity layer beneath the records your AIMS produces.
Frequently asked questions.
01What is ISO 42001 and who needs it?
02How do I comply with the ISO 42001 AI management system standard?
03What evidence does an ISO 42001 auditor expect to see?
04How many controls are in ISO 42001 Annex A?
05What is the difference between ISO 42001 and the EU AI Act?
06Is ISO 42001 certification mandatory?
07Does ar.io provide ISO 42001 certification?
Build records your auditor can verify independently.
Policies are the visible work. The records behind them are what an auditor tests, and with ar.io they are independently verifiable, so an auditor or customer can confirm they are unaltered without taking your word for it. Anchor your impact assessments, event logs, and internal-audit records before your next surveillance review.