Skip to navigation Skip to main content Skip to footer

Adelard supports ONR research on safe and assured AI adoption in the UK nuclear sector

 

Adelard is pleased to have supported the Office for Nuclear Regulation (ONR) with the development of two research reports examining the use of Artificial Intelligence and Machine Learning (AI/ML) systems within the UK civil nuclear industry. The reports are: 

  • Characterisation of safety of AI technologies and AI framework 

  • Route map for regulating and deploying AI/ML systems within the UK civil nuclear industry 

AI and Machine Learning have the potential to deliver significant benefits for the nuclear industry, including improved efficiency, enhanced decision‑making, and the reduction of risk to people and the environment. However, the adoption of these technologies also raises important questions around assurance, uncertainty, and long‑term safety. 

Together, the reports explore how AI/ML could be safely enabled within nuclear operations, providing a structured foundation for regulators, duty holders and system developers. Key topics addressed include: 

  • Characterising AI applications in nuclear operations - Identifying where and how AI could be used across the nuclear system lifecycle, from advisory and analytical tools to more autonomous systems, and the implications for safety. 
  • Levels of autonomy and consequence - Exploring how increasing autonomy changes risk, influences assurance strategies, and affects the role of human oversight. 
  • Assurance challenges unique to AI - Including data‑driven behaviour, limited explainability, verification and validation constraints, and the impact of system change over time. 
  • Risk modelling for AI‑modulated systems - Considering how failure detectability, consequence and recoverability shape deployment and regulatory decisions.
  • Human factors and sociotechnical considerations - Addressing trust, over‑reliance, skill fade, and the need for interfaces and governance that support effective supervision.

 

AI‑Modulated Systems: a system‑level perspective 

A central theme across both reports is the concept of AI‑Modulated Systems (AMSs). Rather than treating AI as a standalone component, an AMS recognises AI as part of a wider sociotechnical system, encompassing people, processes, data, system architecture, development and training activities, and organisational governance. Together, these elements shape overall system behaviour and risk. This system‑level perspective underpins the assurance approach set out in the reports, recognising that the behaviour and risks associated with AI cannot be fully understood by analysing algorithms in isolation. 

 

Route map for regulation and assurance 

Building on Adelard’s earlier research for ONR, the updated route map defines the core elements of a regulatory and assurance framework for AI/ML systems in the nuclear sector. 

It sets out: 

  • Priority research areas 

  • Phased programme of work spanning immediate, short-term and long‑term activities 

  • Capability‑building initiatives to support effective deployment and regulation 

The route map explicitly recognises both the current maturity of AI technologies and the pace at which they are developing, providing a basis for proportionate and adaptable regulatory oversight. 

 

Supporting ONR’s mission 

Adelard is proud to support ONR in enabling innovation by contributing to guidance that helps nuclear stakeholders adopt AI safely, securely and with confidence. 

This work reflects our ongoing commitment to robust, holistic system‑level assurance and to supporting the responsible adoption of novel technologies across safety‑critical industries.