The International RegLab Project has released its first report of the first RegLab cycle, marking a significant milestone in collaborative efforts to explore how innovative technologies, particularly artificial intelligence (AI), can be effectively, safely and transparently integrated into the nuclear sector.
RegLab is a “sandboxing” activity that brings together technologists, operators and regulators to collectively examine how emerging technologies may progress from concept to deployment. Building on experience from regulatory sandboxing in other sectors, such as finance, aviation and medicine, and an earlier international pilot work supported by Electric Power Research Institute (EPRI), the International Atomic Energy Agency's ISOP initiative, and the Canadian Nuclear Society in 2024, the NEA RegLab Project is tailored to address nuclear sector challenges and regulatory expectations.
The newly released report summarises insights from the first full RegLab cycle (RegLab #1), which focused on the use of AI technologies in real-time monitoring of nuclear power plant data to identify operational inconsistencies.
Key findings
AI use case: Real‑time monitoring for anomaly detection
RegLab #1 explored a representative AI application designed to detect anomalies in real-time operational data. Participants from regulatory bodies, industry and the technology community noted the potential benefits of such systems, such as improved safety margins, early detection of deviations and the possibility of reducing operational costs.
Two major challenges emerged clearly across discussions:
- AI explainability
While explainable AI is essential for supporting safety justification, participants concluded that explainability alone is insufficient for applications with high safety implications. The report emphasises that defence-in-depth measures remain foundational. AI systems must be able to “show their working” with quantifiable, auditable justifications to support regulatory confidence. - Data assurance
Robust data assurance was identified as a critical enabler for any AI-supported safety case. Participants emphasised the need for high quality, well governed, representative datasets, not merely data availability, to establish credible technical justification.
Evaluation of the RegLab approach
Participants reported that the RegLab structure was logical, inclusive and effective in fostering constructive dialogue. The iterative development of a hypothetical case encouraged stakeholders to explore issues from multiple viewpoints, including those of regulators, operators and technology developers.
Recommendations for future work
Based on the gaps identified, the report recommends establishing working groups to develop good practice guidance for an AI nuclear assurance framework. Priority areas include:
- standards for AI verification and validation (V&V);
- clarification of practical boundaries for AI applications;
- approaches to managing residual risks using defence-in‑-‑depth measures;
- enhanced training and competency development for both AI developers and nuclear end-users; and
- harmonisation of metadata structures and taxonomies to support cross-industry consistency.
These actions are intended to support both innovation and regulatory robustness, ensuring that AI technologies can be deployed safely and effectively across diverse nuclear applications.
International collaboration
Implemented under the auspices of the Nuclear Energy Agency (NEA) and in co-operation with the Electric Power Research Institute (EPRI) and the International Atomic Energy Agency (IAEA) International Network on Innovation to Support Operating Nuclear Power Plants (ISOP), the initiative is supported by regulatory bodies and technical support organisations, including:
- Canadian Nuclear Safety Commission (CNSC)
- Autorité de Sûreté Nucléaire et de Radioprotection (ASNR), France
- Nuclear Regulation Authority, Japan
- Korea Atomic Energy Research Institute (KAERI)
- Consejo de Seguridad Nuclear (CSN), Spain
- Office for Nuclear Regulation (ONR), United Kingdom
- US Nuclear Regulatory Commission (NRC)
Looking ahead
As AI technologies advance rapidly, the RegLab initiative is expected to play a growing role in helping the international nuclear community understand both the opportunities and the challenges associated with their application. The insights from RegLab #1 provide an early foundation for harmonised approaches and open a new chapter in internationally co-ordinated regulatory innovation.
The full report is available on the Ennuvo website.
