If you have difficulty in submitting comments on draft standards you can use a commenting template and email it to admin.start@bsigroup.com. The commenting template can be found here.
Standards Development
This document specifies the requirements and guidance for achieving appropriate level of explainability in applying artificial intelligence (AI) in the nuclear sector. This document doesn’t apply to instrumentation and control systems important to safety. It covers both nuclear applications based on open -source AI algorithms, as well as AI algorithms that are developed specifically for nuclear applications. It is applicable to both nuclear power plants and other nuclear facilities.
Other aspects of ensuring required performance of AI techniques for nuclear applications are beyond the scope of this document.
Artificial Intelligence (AI) has been widely applied across various fields of industries, including the nuclear industry. Many countries and international organizations are actively exploring and advancing the potential applications of AI techniques in nuclear facilities. However, as many AI techniques, in particular deep learning models, are different from traditional software and to some extent represent black -box characteristics. This leads to opacity and poor explainability, which restricts its widespread adoption in the nuclear sector, due to the stringent requirements of safety and reliability in the nuclear sector. The International Atomic Energy Agency (IAEA) and many nuclear regulatory bodies have identified explainability as one of the priority top ics to be resolved to facilitate adoption of AI techniques in the nuclear industry. IEC SC45A published IEC TR 63468 “Nuclear facilities - Instrumentation and control, and electrical power systems - Artificial Intelligence applications” in 2023, which explicitly identified explainability equirements as an important horizontal standard to be developed at the second tier under the framework of a four - tier standard architecture.
Up to now, several standards related to explainability have been published in the general AI field, such as ISO/IEC TS 6254:2025 “Information Technology - Artificial Intelligence - Objectives and Approaches for Explainability and Interpretability of Machine Learning (ML) Models and Artificial Intelligence (AI) Systems” and IEEE 2894:2024 “IEEE Guide for an Architectural Framework for Explainable Artificial Intelligence”. However, such standards focus on general AI technologies and do not fully take into account the specific requirements of AI applications in nuclear facility scenarios. In particular, there are differences between their life cycle models and that for the nuclear sector as already established in the IEC standards and widely adopted world-wide. Therefore, it is imperative to develop more specific guidance within the framework of the IEC SC45A standards, respecting the unique needs of nuclear facility applications and taking into account diverse stakeholder interests.
Required form fields are indicated by an asterisk (*) character.
You are now following this standard. Weekly digest emails will be sent to update you on the following activities:
You can manage your follow preferences from your Account. Please check your mailbox junk folder if you don't receive the weekly email.
You have successfully unsubscribed from weekly updates for this standard.
Comment by: