Overview
ISO/IEC TS 42119-2:2025 is a Technical Specification developed by ISO and IEC to provide requirements and guidance for testing Artificial Intelligence (AI) systems. Building upon the established ISO/IEC/IEEE 29119 software testing standards, this document offers tailored approaches to address the unique complexities and risks of AI systems throughout their lifecycle. By incorporating a risk-based testing framework, ISO/IEC TS 42119-2:2025 helps organizations systematically identify, assess, and mitigate potential risks associated with AI, ensuring robust test practices that are aligned to stakeholder needs, industry quality benchmarks, and evolving regulatory expectations.
Key Topics
-
Risk-based Testing for AI
This standard emphasizes a risk-based approach, treating risk identification and prioritization as fundamental to shaping test strategies for AI systems. By integrating ISO/IEC/IEEE 29119 processes, organizations can determine the most suitable test techniques, levels, and documentation requirements tailored to AI-related risks.
-
Integration with the ISO/IEC/IEEE 29119 Series
The document details how to apply the software test processes, documentation templates, and test design techniques from ISO/IEC/IEEE 29119-2, -3, and -4 to AI systems and their components. This is essential for aligning AI system testing with international software engineering standards.
-
AI-specific Testing Considerations
ISO/IEC TS 42119-2:2025 covers specialized AI system aspects, including coverage of AI model validation, testing of data quality, and static analysis of knowledge engineering systems. It guides the application of both manual and automated test approaches within the context of AI technologies.
-
AI System Lifecycle and Testing Processes
The document outlines the relationship between the AI system lifecycle-spanning design, development, deployment, and retirement-and corresponding testing processes. Test practices should adapt to each lifecycle stage, ensuring continuous oversight and adaptation as AI systems evolve.
Applications
Organizations implementing AI systems in domains such as finance, healthcare, autonomous vehicles, and industrial automation benefit from adopting ISO/IEC TS 42119-2:2025 by:
-
Improving AI Model Reliability
Utilizing risk-based testing and comprehensive documentation to identify and reduce errors, biases, and vulnerabilities in AI models.
-
Enhancing Transparency and Accountability
Ensuring all stakeholders are identified and engaged during the testing process, which helps meet compliance, ethical, and quality demands.
-
Facilitating Regulatory Compliance
Supporting adherence to national and international regulations by providing a standardized approach to AI testing, bolstering trust from users and authorities.
-
Streamlining AI Project Management
By following clearly defined processes for test management, documentation, and coverage, organizations can efficiently plan, execute, and report on AI system testing activities.
-
Mitigating AI-Related Risks
Applying structured risk assessment methods to address unique threats like bias, data quality issues, and concept drift found in AI-based solutions.
Related Standards
Organizations applying ISO/IEC TS 42119-2:2025 should also consider the following standards:
-
ISO/IEC/IEEE 29119 series
International standards for software testing, including test processes, test documentation, and test design techniques, which form the foundation for AI system testing in this specification.
-
ISO/IEC 22989
Definitions and concepts for AI systems, which help inform the functional, architectural, and risk-based perspectives in testing.
-
ISO/IEC 23894
Guidelines for risk management within AI, supporting structured risk identification and assessment.
-
ISO/IEC 20246
Provides guidance on reviews, inspections, and technical documentation validation, applicable to AI system development and testing activities.
-
ISO/IEC TS 42119-3
Focuses on verification and validation analysis of AI systems.
-
ISO/IEC TS 42119-7 and -8
Offer specific guidance for red teaming assessments and prompt-based generative AI systems.
By adhering to ISO/IEC TS 42119-2:2025, organizations can ensure that the testing of their AI systems is systematic, risk-focused, and grounded in recognized international best practices.