ISO/IEC 5338: AI System Life Cycle Processes Explained
Standards & Compliance

ISO/IEC 5338

ISO/IEC 5338:2023 is an ISO/IEC JTC 1/SC 42 standard that defines life cycle processes for organizations engineering AI systems. It extends the systems and software engineering life cycle models in ISO/IEC 15288 and ISO/IEC/IEEE 12207 with AI-specific process outcomes covering concept, development, production, utilization, support, and retirement stages. It specifies what process outcomes are required, not how to implement them, and it does not include certification criteria.

ISO/IEC 5338 at a Glance

AttributeDetail
Developed byISO/IEC JTC 1/SC 42, the joint AI standardization committee
ExtendsISO/IEC 15288 and ISO/IEC/IEEE 12207 life cycle models
StructureProcess outcomes, not prescribed techniques or tools
CertifiableNo, unlike ISO/IEC 42001, which is auditable
Companion standardsISO/IEC 22989 (terminology), ISO/IEC 23894 (risk)

What ISO/IEC 5338 Defines

ISO/IEC 5338:2023 is published by ISO/IEC JTC 1/SC 42, the joint technical committee responsible for AI standardization. It defines life cycle processes for organizations that engineer AI systems, extending the established systems and software engineering life cycle models in ISO/IEC 15288 and ISO/IEC/IEEE 12207 with process outcomes specific to AI.

The standard specifies what a life cycle process must achieve; it does not prescribe how those outcomes should be implemented, and it does not include certification or conformity assessment criteria.

Scope and Life Cycle Stages

ISO/IEC 5338 organizes an AI system's life cycle into six stages: concept, development, production, utilization, support, and retirement. The retirement stage covers decommissioning of AI models once they are no longer in use, an outcome that many organizations' existing software life cycle practices do not explicitly address.

Because the standard defines outcomes rather than methods, organizations can map it onto existing S