ISO IEC TS 42119-2-2025 PDF

Full title and description

ISO/IEC TS 42119-2:2025 — Artificial intelligence — Testing of AI — Part 2: Overview of testing AI systems. This technical specification provides an overview and guidance for applying established software testing standards and techniques to AI systems, with a focus on risk-informed selection of test practices and the mapping of testing activities to AI lifecycle stages.

Abstract

This document provides requirements and guidance on the application of the ISO/IEC/IEEE 29119 series to the testing of AI systems. It follows a risk‑based approach that uses risks associated with AI systems, their development and maintenance, to identify suitable test practices, approaches and techniques applicable to AI systems and their components. Where testing practices are already covered in the ISO/IEC/IEEE 29119 series, the specification gives additional detail on applying those practices in the AI context and describes how AI/ML assessment metrics can be used with common testing techniques.

General information

• Status: Published
• Publication date: 3 November 2025 (Edition 1, 2025)
• Publisher: Joint ISO/IEC publication (prepared by ISO/IEC JTC 1/SC 42)
• ICS / categories: 35.020 (Information technology)
• Edition / version: Edition 1.0 (2025)
• Number of pages: 34

Key registration and publication details are recorded on the ISO and IEC publication records.

Scope

The scope covers the overview of testing techniques applicable to AI systems across the AI system life cycle, including how techniques described in ISO/IEC/IEEE 29119 (notably test processes and test techniques) apply to AI and ML components. The specification maps testing processes to verification and validation stages defined in AI lifecycle guidance (for example ISO/IEC 22989) and describes the use of AI/ML assessment metrics in test activities. The document is technology‑agnostic and intended to be applicable to a wide range of AI system types and architectures.

Key topics and requirements

• Application of ISO/IEC/IEEE 29119 test processes and techniques to AI systems (adaptation and additional guidance for AI contexts).
• Risk-based selection of test practices tied to identified AI system risks and lifecycle stages.
• Mapping of testing activities to verification and validation phases of the AI lifecycle (including model development, deployment and maintenance).
• Guidance on using AI/ML assessment metrics within standard test techniques and on interpreting metric results for verification/validation.
• Recommendations for test planning, test data considerations, environment and reproducibility concerns specific to AI/ML.

These topics align testing processes and documentation with AI-specific concerns such as dataset bias, model drift, performance variability and explainability.

Typical use and users

Intended users include test engineers and QA teams working on AI/ML systems, software developers integrating AI components, system validators, regulators and conformity assessment bodies, product managers responsible for AI quality, and organizations establishing test processes for AI products. The specification is used to inform test strategy, test planning and selection of techniques appropriate to identified AI risks.

Related standards

Closely related and commonly referenced standards and projects include the ISO/IEC/IEEE 29119 software testing series (test processes and test techniques), ISO/IEC 22989 (AI concepts and terminology / AI lifecycle), and other parts of the ISO/IEC 42119 series (related parts and work items such as Part 1 and subsequent verification/validation parts and red‑teaming guidance under development). These documents together provide terminology, lifecycle models, testing process models and specific techniques that the TS 42119‑2 text adapts or references for AI contexts.

Keywords

AI testing, testing of AI, ISO/IEC TS 42119, software testing, ISO/IEC/IEEE 29119, risk‑based testing, verification and validation, AI lifecycle, ML metrics, test techniques, reproducibility, test planning.

FAQ

Q: What is this standard?

A: ISO/IEC TS 42119-2:2025 is a Technical Specification that provides an overview and guidance for testing AI systems by adapting and applying the ISO/IEC/IEEE 29119 software testing series to AI contexts.

Q: What does it cover?

A: It covers a risk‑based overview of test practices, mapping of test processes to AI lifecycle verification and validation stages, the use of AI/ML assessment metrics with common test techniques, and guidance on selecting appropriate test approaches for AI system components.

Q: Who typically uses it?

A: Test and QA engineers, developers and architects working with AI/ML systems, product owners, regulators and conformity assessment bodies, and organisations establishing or auditing AI testing processes.

Q: Is it current or superseded?

A: As published, the document is a current Technical Specification issued in 2025 (Edition 1). Check national adoption records or the issuing bodies for any later amendments or related parts published after 3 November 2025.

Q: Is it part of a series?

A: Yes — TS 42119‑2 is part of the broader ISO/IEC 42119 work on testing of AI, which includes other parts and related project work items (e.g., planned or in‑progress parts for verification/validation, red‑teaming and Part 1 project items). Users typically apply TS 42119‑2 in conjunction with ISO/IEC/IEEE 29119 and ISO/IEC 22989.

Q: What are the key keywords?

A: AI testing, risk‑based testing, ISO/IEC 42119, 29119, verification, validation, ML metrics, test techniques, AI lifecycle, reproducibility.