Standard for Safety for AI-Based Products
Scope:
The scope of this Standard is for AI-based products, which can be AI/ML applications (software products with AI-based or ML-based capabilities) such as a fault detection ML model or a Large Language Model (LLM) based chatbot for customer service; or existing products with added AI-based capabilities or components.
This Standard is intended to specify a horizontal set of safety requirements for various AI-based products, and it is to be used in conjunction with other existing (vertical) product safety standards.
Not all the requirements specified in this document apply to every AI-based product. Depending on the particular AI application or product, only a subset of these requirements may be relevant.
AI-based products and services to be covered by these requirements include but are not limited to appliances, autonomous driving, batteries, consumer electronics, furniture, HVAC/R, household and industrial cleaners, lighting, wire and cable, robotics, and medical devices.
Relevant AI application domains as selected from ISO/IEC TR 24030:2024 include but are not limited to agriculture, construction, defense, education, energy, legal, manufacturing, retail, security, and transportation.
Relevant ISO/IEC standards related to artificial intelligence are included as normative references.
Project need:
The landscape of Artificial Intelligence (AI) is evolving with unprecedented speed, introducing AI-based functionalities into a broad spectrum of consumer and industrial products. This evolution is marked by significant milestones such as the advent of Machine Learning (ML), the resurgence of Deep Learning (DL) technologies and emergence of Generative AI. These advancements have found applications across various sectors, enhancing wellness, convenience, productivity, efficiency, and innovation, while also posing new challenges in terms of ethical considerations and safety risks.
This project is to develop a joint Canada-United States National Standard using a comprehensive framework for evaluating the safety of AI-Based products. This framework will outline the necessary requirements, procedures, and guidelines to ensure these products are safe to use and align with the highest standards of AI safety, regulations, and industry best practices. It is aligned with existing and upcoming standards and methodologies defined by recognized standards organizations such as ISO/IEC JTC 1/SC 42 and IEEE-SA and regulatory bodies such as the U.S. AI Executive Order and the EU AI Act, ensuring a holistic and rigorous evaluation process. This standard is also intended to bridge the gap between rapid technological advancement and the development of corresponding safety standards, ensuring AI-based products are not only innovative and efficient but also safe, ethical, and trustworthy. Establishing AI system safety is vital for building public trust and ensuring the responsible adoption of beneficial AI technologies.
Key stakeholders in this standardization activity include producers of AI-based software, consumers that use AI-based products, and manufacturers of products that use AI-based software. This joint ANSI/CAN/UL Standard will support certification programs in both Canada and the U.S., facilitating cross-border consistency and ensuring alignment with industry best practices. Its maintenance process will enable rapid updates to address emerging needs, ensuring the standard evolves alongside advances in technology and market demands.
Note: The information provided above was obtained by the Standards Council of Canada (SCC) and is provided as part of a centralized, transparent notification system for new standards development. The system allows SCC-accredited Standards Development Organizations (SDOs), and members of the public, to be informed of new work in Canadian standards development, and allows SCC-accredited SDOs to identify and resolve potential duplication of standards and effort.
Individual SDOs are responsible for the content and accuracy of the information presented here. The text is presented in the language in which it was provided to SCC.