Technical Specification
Track updates
SA TS ISO/IEC 12791:2025
[Current]SA TS ISO/IEC 12791:2025 identically adopts ISO/IEC TS 12791:2024, which describe how to address unwanted bias in AI systems that use machine learning to conduct classification and regression tasks. Provides mitigation techniques that can be applied throughout the AI system life cycle in order to treat unwanted bias. Applicable to all types and sizes of organization.
Published: 09/05/2025
Pages: 25
Table of contents
Cited references
Content history
Table of contents
Header
About this publication
Preface
Foreword
Introduction
1 Scope
2 Normative references
3 Terms and definitions
3.1 General
3.2 Artificial intelligence
3.3 Bias
3.4 Testing
4 Abbreviated terms
5 Treating unwanted bias in the AI system life cycle
5.1 Inception
5.1.1 Stakeholder identification
5.1.2 Stakeholder needs and requirements definition
5.1.3 Procurement
5.1.4 Data sources
5.1.5 Integration with risk management
5.1.6 Acceptance criteria
5.2 Design and development
5.2.1 Feature representation
5.2.2 Metadata sufficiency
5.2.3 Data annotations
5.2.4 Adjusting data
5.2.5 Methods for managing identified risks
5.3 Verification and validation
5.3.1 General
5.3.2 Static testing of data used in development
5.3.3 Dynamic testing
5.4 Re-evaluation, continuous validation, operations and monitoring
5.4.1 General
5.4.2 External change
5.5 Disposal
6 Techniques to address unwanted bias
6.1 General
6.2 Algorithmic and training techniques
6.2.1 General
6.2.2 Pre-trained models
6.3 Data techniques
7 Handling bias in a distributed AI system life cycle
Annex A
Annex B
B.1 Workers' needs
B.2 Vulnerable consumers
Bibliography
Cited references in this standard
[Current]
Artificial intelligence - Data quality for analytics and machine learning (ML) - Part 4: Data quality process framework
[Current]
Information technology - Artificial intelligence - Artificial intelligence concepts and terminology
[Current]
Software and systems engineering - Software testing - Part 3: Test documentation