Global Standards for Responsible AI Development

The standard establishes requirements for organizations to create, implement, and maintain systems for responsible governance of AI throughout its lifecycle.

ISO/IEC 22989 provides standardized definitions and terminology for AI concepts, enabling consistent communication across technical teams, business leaders, and regulatory bodies.

ISO/IEC 23894:2023 adapts traditional risk management principles to help organizations identify, assess, and manage AI-specific risks throughout the system lifecycle.

Provides technical guidance for identifying, measuring, and reducing human, data, and engineering biases throughout the AI system lifecycle.

Provides a framework for understanding and establishing trustworthiness in AI systems through transparency, explainability, controllability, and risk assessment approaches.

ISO/IEC TR 24029-1:2021 provides methods for assessing neural networks' ability to maintain performance when exposed to unexpected inputs, adversarial attacks, or environmental changes.

ISO/IEC 23053 establishes a framework for describing machine learning-based AI systems by defining essential components, functions, and common terminology for development and governance.

ISO/IEC 25059:2023 extends software quality standards to evaluate AI-specific characteristics including learning capabilities, probabilistic reasoning, explainability, fairness, and incomplete data handling.

ISO/IEC 27001 provides privacy controls for protecting personal data, including data processed by AI systems requiring privacy compliance.

Establishes requirements for managing information security risks through systematic controls, serving as a foundation for securing AI systems and data.

Provides guidance for governing bodies to establish oversight, policies, and risk management for responsible AI deployment aligned with organizational objectives.

An IT governance framework adapted for managing AI systems through structured processes, risk management, and performance measurement aligned with business objectives.

ISO/IEC 27002 provides detailed security control implementation guidance for protecting information assets, including AI systems, data, and models from cybersecurity threats.

ISO/IEC 25012 defines fifteen measurable data quality characteristics that provide a critical foundation for AI system reliability, accuracy, and trustworthiness.

The foundational framework, terminology, and examples for ensuring data quality in machine learning and analytics applications.

Defines fourteen standardized data quality metrics and assessment methodologies for evaluating and measuring data quality in AI and machine learning systems.

Establishes data quality requirements and guidelines for AI systems, focusing on training, validation, and test datasets throughout the lifecycle.

Provides a process framework for ensuring data quality throughout the lifecycle of training and evaluation in machine learning and analytics applications.

Establishes a governance framework for directing, overseeing, and controlling data quality measures throughout the lifecycle of analytics and machine learning systems.

Provides technical specifications for developing responsible AI systems with emphasis on safety, ethics, trustworthiness, and human oversight in engineering applications.

Provides a values-neutral framework for organizations to identify and address ethical and societal concerns throughout the AI lifecycle.

Establishes quality management system requirements that AI organizations can apply to ensure consistent processes, risk management, and continuous improvement in AI development.

Provides a business continuity management framework that organizations can apply to ensure AI systems remain operational during disruptions and incidents.

ISO 14001 for AI integrates environmental management systems with AI operations, helping organizations reduce environmental impact while optimizing AI performance.

IEC 38500 extends corporate IT governance principles to AI systems, providing organizations with a framework for responsible oversight, strategy alignment, and risk management.

Provides a data quality assessment framework originally for geographic information that organizations can adapt for evaluating AI training data quality.

The standard establishes safety and security requirements for health software products operating on general computing platforms, including AI-powered medical software applications.

Provides technical guidance for safely integrating artificial intelligence components into systems where failures could harm people, property, or environment.

Defines standardized data quality measures and assessment methodologies specifically designed for artificial intelligence and machine learning systems.

ISO/IEC 8183:2023 establishes a ten-stage framework for managing data throughout AI system lifecycles from conception to decommissioning.

ISO/IEC 42005 provides guidance for organizations to assess potential impacts of AI systems on individuals and society throughout the system lifecycle.

Learn about the latest AI standards in development and global efforts to harmonize them across industries and regions for safer AI use.