ISO AI Governance Overview

Overview

The International Organization for Standardization (ISO) is an independent, non-governmental international organization that plays a critical role in developing and publishing international standards. In the realm of Artificial Intelligence (AI), ISO, in collaboration with the International Electrotechnical Commission (IEC), spearheads governance activities through their joint technical committee, ISO/IEC JTC 1/SC 42 (Artificial Intelligence). This committee is dedicated to creating comprehensive international standards, technical reports, and technical specifications that ensure the quality, safety, efficiency, and trustworthiness of AI systems globally, thereby fostering responsible development and deployment across various industries.

AI Governance Mandate

ISO/IEC JTC 1/SC 42 serves as the central point for AI standardization within both ISO and IEC, with a broad mandate to address the entire AI ecosystem. Its extensive program of work encompasses foundational AI standards, data-related standards for AI, big data and analytics, AI trustworthiness, use cases and applications, governance implications of AI, computational approaches, testing of AI systems, and ethical and societal concerns. The committee adopts an ecosystem approach, integrating regulatory, business, societal, and ethical perspectives to develop comprehensive standards that are universally applicable. This mandate is crucial for ensuring that AI technologies are developed and utilized responsibly, ethically, and effectively, aligning with international best practices and promoting a human-centric approach to AI.

Key AI Instruments

ISO has developed a suite of key standards and technical reports that form the backbone of its AI governance framework:

• ISO/IEC 42001:2023 – Artificial Intelligence Management System (AIMS): This is the foundational and certifiable standard, providing a framework for organizations to establish, implement, maintain, and continually improve an AIMS. It addresses ethical considerations, transparency, AI system impact assessment, risk management, lifecycle management, and third-party supplier oversight, promoting responsible and ethical AI practices.
• ISO/IEC 23894:2023 – Information technology — Artificial intelligence — Guidance on risk management: This standard offers guidance on identifying, assessing, and addressing key sources of risk associated with AI systems, including those related to data and personnel.
• ISO/IEC 22989:2022 – Information technology — Artificial intelligence — Artificial intelligence concepts and terminology: This document provides a common language and understanding of fundamental AI concepts and terminology, which is essential for consistent governance and development across the AI lifecycle.
• ISO/IEC TR 24028:2020 – Information technology — Artificial intelligence — Overview of trustworthiness in artificial intelligence: This technical report surveys various topics related to building trustworthiness in AI systems, including approaches to establish trust through transparency, explainability, and controllability, as well as identifying and mitigating engineering pitfalls and associated threats.
• ISO/IEC TR 24027:2021 – Information technology — Artificial intelligence (AI) — Bias in AI systems and AI aided decision making: This technical report describes measurement techniques and methods for analyzing, assessing, and addressing bias in AI systems and AI-aided decision-making processes.
• ISO/IEC TS 8200:2024 – Information technology — Artificial intelligence — Controllability of automated artificial intelligence systems: This technical specification outlines a framework with principles, characteristics, and approaches designed to enhance the controllability, reliability, and alignment with human values for automated AI systems.

Implementation Mechanisms

ISO standards are developed through a voluntary, consensus-based process, fostering global collaboration among experts. The adoption of these international standards, particularly ISO/IEC 42001, allows organizations to demonstrate a robust commitment to ethical AI practices and to align with both international best practices and evolving regulatory requirements worldwide. ISO/IEC 42001 is designed to be certifiable, meaning organizations can undergo independent third-party assessments to verify compliance with its requirements, thereby enhancing trust with stakeholders and customers. National standards bodies globally play a key role in adopting and promoting these international standards within their respective jurisdictions.

Member Engagement

ISO operates as an independent, non-governmental organization, drawing expertise from its 172 member countries. ISO/IEC JTC 1/SC 42 actively engages experts from over 60 participating countries, ensuring a diverse and inclusive development process for AI standards. The "one country, one vote" system within ISO ensures that standards reflect a broad range of global perspectives and needs, promoting international alignment and cooperation. Furthermore, the committee organizes bi-annual AI workshops, bringing together innovators, researchers, developers, and other stakeholders from various sectors to facilitate dialogue on emerging trends, technologies, requirements, and the role of standards in the AI ecosystem.

Current Focus Areas

The current program of work for ISO/IEC JTC 1/SC 42 is extensive, encompassing the development of foundational AI standards, data standards pertinent to AI, big data and analytics, AI trustworthiness, various use cases and applications, the governance implications of AI, computational approaches, testing methodologies for AI systems, and addressing ethical and societal concerns. A significant emphasis is placed on creating international standards that facilitate trustworthy digital transformation and tackle emerging challenges such as explainability, controllability, robustness, and the mitigation of bias in AI systems.

Future Directions

ISO's future directions in AI governance are poised to continuously expand its suite of standards to keep pace with rapid technological advancements and evolving societal demands. This trajectory includes the ongoing development of technical reports into full international standards, the refinement of existing standards, and enhanced collaboration with other ISO, IEC, and JTC 1 committees on both horizontal technologies and vertical applications of AI. The overarching goal remains to provide a comprehensive, adaptable, and globally relevant framework that supports innovation while ensuring the responsible, ethical, and safe deployment of AI worldwide. Insights gleaned from the bi-annual workshops will continue to inform the strategic planning and road-mapping of future work items.

Resources

Standards Tracked

The 14 ISO and ISO/IEC AI instruments currently tracked on this site.