ISO 42001 certification: AI management systems

ISO/IEC 42001 is the first international standard for artificial intelligence management systems. It sets out how organisations develop, provide and use AI systems responsibly.

Small measuring station with a solar panel at the edge of a glacier
Anemometer and wind vane on a mast at the summit

ISO/IEC 42001 covers an AI system from planning through development, verification and operation to decommissioning.

Phases in the life cycle of an AI system
6
Ways of working with AI: develop, provide, use
3
until your quotation
24 h

An AI system from concept to decommissioning.

Requirements, evidence and responsibilities are defined for each phase.

123456
  1. ConceptDefine purpose and area of use.
  2. DevelopmentBuild data and model.
  3. VerificationCheck results before deployment.
  4. OperationUse the system under human oversight.
  5. MonitoringContinuously observe results and impacts.
  6. DecommissioningRetire the system in an orderly manner.

The impact assessment accompanies the entire arc.

The standard is aimed at all organisations that develop, provide or use AI systems, regardless of size and sector. It follows the same structure as ISO 9001 and ISO/IEC 27001 and can therefore be integrated into an existing management system.

With an AI management system in accordance with ISO/IEC 42001, you show customers, partners and authorities that the use of AI in your organisation is governed, documented and monitored.

  1. Governance and accountability

    The standard requires an AI policy, clear roles and objectives for the use of AI. Top management defines what AI is and is not used for in the organisation and provides the necessary resources and competences.

    This creates a framework in which specialist departments, IT and compliance work to the same rules.

  2. Assessing risks and impacts

    At the core of the standard are the risk assessment and the impact assessment for AI systems. These evaluate the effects a system may have on individuals, groups and society, for example through incorrect results, bias or a lack of traceability.

    You derive measures from this assessment. The annex to the standard lists controls for this purpose, for example on data quality, transparency and human oversight.

  3. The life cycle of AI systems

    ISO/IEC 42001 considers an AI system across its entire life cycle, from planning through development, testing and operation to decommissioning. Requirements, evidence and responsibilities are defined for each phase.

    Internal audits and the management review ensure that the system is continually improved and keeps pace with new legal requirements.

Rows of steel avalanche barriers on a steep slope

Who needs an AI management system.

  • You develop AI systems

    You want to define requirements and evidence for every phase.

  • You provide AI

    Customers, partners and authorities ask how its use is governed and monitored.

  • You use AI in your operations

    Management wants to define what AI is used for and what it is not used for.

  • You hold ISO 9001 or ISO/IEC 27001

    ISO/IEC 42001 follows the same structure and fits into your existing system.

How your AI management system is audited.

  1. Application

    You name the AI systems within the scope and your role: development, provision or use. SICE prepares the quotation.

  2. Stage 1

    SICE reviews the documentation: AI policy, roles, risk assessment and impact assessment.

  3. Stage 2

    On-site audit. SICE examines the controls on selected AI systems, for example on data quality, transparency and human oversight.

  4. Certificate

    Following a positive decision, SICE issues the certificate to ISO/IEC 42001.

  5. Surveillance

    New AI systems and new legal requirements are included in the surveillance audits.

Questions on the use of AI.

Does ISO/IEC 42001 also apply if we only use AI?

Yes. The standard is aimed at all organisations that develop, provide or use AI systems, regardless of size and sector.

What is an impact assessment for AI systems?

It assesses the impact a system may have on individuals, groups and society, for example through incorrect results, bias or a lack of traceability.

Which controls does the standard specify?

The annex to the standard lists controls, for example on data quality, transparency and human oversight.

Who decides what AI is used for?

Top management. It decides what AI is and is not used for in the organisation and provides the necessary resources and competences.

Does the standard fit our existing management system?

Yes. It follows the same structure as ISO 9001 and ISO/IEC 27001 and can be integrated into an existing management system.

Small measuring station with a solar panel at the edge of a glacier

Let us talk about AI in your organisation.

In the free initial consultation, we clarify which AI systems belong within the scope.