The International Atomic Energy Agency (IAEA) is known for collaborating with designated institutions around the world to promote the peaceful use of nuclear technologies and seeking assistance with research, development, and training. The focus is on nuclear science, technologies, and their safe and secure applications around the world.
The IAEA has designated the 73rd Centre in 2024 for Science of Information at Purdue University in the United States of America as the first IAEA Collaborating Centre to support the Agency’s activities on Artificial Intelligence (AI) for nuclear power applications, including reactor design, plant operation, training, and education.
The five-year Collaborating Centre agreement is designed to seek support for programmatic activities and knowledge sharing advancements and innovation in AI for nuclear power. This includes:
- Agency initiatives on benchmark exercises for developing confidence;
- Community-wide acceptance of Al technology for nuclear power, establishing a “Benchmarking Hub” for coordination and data management; and
- Other activities relevant to the development and assessment of Al technologies in collaboration with IAEA Member States.
Incidentally, there were 180 IAEA members around the world as of 15 November 2024.
Furthermore, the Collaboration Centre Agreement is part of recent IAEA efforts to strengthen support to countries interested in using AI for nuclear science and technology.
The timing for the agreement was appropriate as it happened right after the designation of the Massachusetts Institute of Technology (MIT) Plasma Science and Fusion Center. It is the first Collaborating Centre focused on accelerating fusion research, with an emphasis on AI applications to advance the IAEA’s AI for Fusion initiative.
As far as computational resources and data analysis is concerned, the nuclear industry has already started to benefit from AI, including machine learning techniques that can streamline nuclear power plant operations and maintenance. AI is also supporting the development of advanced nuclear power technologies such as small modular reactors (SMRs).
In 2021, the IAEA hosted a Technical Meeting on Artificial Intelligence for Nuclear Technology and Applications aimed at providing an international, cross-cutting forum to discuss and foster cooperation in:
- Nuclear Science;
- Technology and Applications;
- Radiation Protection; and
- Nuclear Security and Safeguard.
The objective was to identify priorities for future activities in these fields and how the IAEA can support their implementation. The meeting — the first of its kind — was also an opportunity to reflect on the ethical considerations of the convergence of artificial intelligence and nuclear technologies.
Consequently, a group of selected professionals conducted a thorough investigation by reviewing the challenges and priorities for future AI activities, including those relevant to nuclear power as well as nuclear science and applications, among others. The IAEA’s International Network on Innovation to Support Operating Nuclear Power Plants (ISOP) was examining the regulatory and technical aspects of AI deployment. Several coordinated research projects related to AI are underway to launch on how AI and other innovative technologies proposed for SMRs can be secured.
Accordingly, a report was published in 2022 by the International Atomic Energy Agency, Vienna with the title: Artificial Intelligence for Accelerating Nuclear Applications, Science and Technology
The report sets out to review the current state of AI/ML (Machine Language) in nuclear science and technology, identify challenges, and outline priorities for future work. It covers:
- Definitions and capabilities of AI methods, including machine learning and deep learning;
- IAEA’s role in promoting safe, ethical, and effective use of AI across its programme areas; and
- A chapter on ethics specifically, which underpins all applications across nuclear domains.
This framing report explains that AI techniques are being explored not as replacements for humans, but as tools to process complex, large datasets, improve modelling, and accelerate discovery or operations where conventional methods are limited.
The report demonstrates that AI and machine learning are powerful enablers across the nuclear sector — accelerating research in nuclear science, optimizing operations in power and radiation protection, enhancing security systems, and strengthening safeguards verification. However, realizing this potential requires careful validation, ethical governance, cross-disciplinary partnerships, and capacity building to ensure AI is safe, effective, and aligned with global nuclear safety and non-proliferation objectives.
The report is structured to address how AI can help accelerate research and practical applications that contribute to health, agriculture, water and environmental science, and fundamental nuclear data:
NUCLEAR APPLICATIONS
This area addresses how AI can help accelerate research and practical applications that contribute to health, agriculture, water and environmental science, and fundamental nuclear data:
| 1 | NUCLEAR APPLICATIONS & NUCLEAR SCIENCE: |
| This area addresses how AI can help accelerate research and practical applications that contribute to health, agriculture, water and environmental science, and fundamental nuclear data; | |
| 1.1 | Human Health: |
| 1.1.1 | AI methods (e.g., deep learning) assist in medical imaging, radiotherapy planning, and dose prediction — improving accuracy and efficiency beyond traditional statistical tools; and |
| 1.1.2 | AI can also support large-scale clinical research and epidemiological datasets; |
| 1.2 | Food & Agriculture: |
| 1.2.1 | AI supports analysis and optimisation in agricultural production systems, food safety verification, and supply chain management when combined with nuclear techniques like isotopic tracing; and |
| 1.3 | Water & Environment: |
| 1.3.1 | AI applied to isotopic and sensor datasets can uncover patterns in hydrology and environmental change that are too complex to derive manually, aiding climate impact assessment and resource planning; and |
| 1.4 | Fundamental Nuclear Data & Physics: |
| 1.4.1 | AI accelerates evaluation and interpolation of nuclear data, and supports theoretical models used in simulation, fusion research, and experiment design — domains where computational complexity is high. |
| IAEA Priorities in Science: Development of FAIR datasets, interdisciplinary workforce training, and shared infrastructure to help scientists and engineers broadly adopt AI methods. | |
| 2 | NUCLEAR POWER: |
| AI methods are considered across the lifecycle of nuclear power technology — including design, operation, maintenance, and safety assessment. | |
| 2.1 | Operations & Monitoring: |
| 2.1.1 | ML models can identify anomalies in sensor data in real time and support predictive maintenance, reducing unplanned downtime and enhancing equipment reliability; |
| 2.2 | Simulation & Optimization: |
| 2.2.1 | Coupling AI with simulators or digital twins (virtual representations of physical systems) can help optimize reactor control, thermal hydraulics, and power output; and |
| 2.3 | Safety & Risk Assessment: |
| 2.3.1 | AI can enhance probabilistic safety assessments by identifying patterns in operational and failure data that are difficult to capture with traditional codes. |
| Challenges Highlighted: | |
| Ensuring explain ability, robustness, and regulatory acceptability of AI — especially when applied to safety-critical systems. | |
| IAEA Priorities in Science: Member State technical meetings and guidance development efforts emphasize developing standards, data sharing, and a stakeholder community that includes regulators, utilities, and technologists. | |
| 3 | RADIATION PROTECTION: |
| While not always segmented as a standalone application category in the core report text, radiation protection is consistently presented as a cross-cutting theme where AI’s analytical capabilities enhance safety: | |
| 3.1 | AI CONTRIBUTION: |
| 3.1.1 | Pattern Recognition in Sensor Data: |
| Machine learning can discern subtle changes in radiation monitoring networks that might indicate unexpected exposures or environmental releases; and | |
| 3.1.2 | Dose Distribution Modelling: |
| AI aids complex model refinement for estimating dose distributions in occupational settings or emergency conditions. | |
| Integration with Safety Systems: The publication draws connections between AI-supported radiation protection and cross-functional safety assessments, reinforcing the need for standards that can safely incorporate these tools into operational practice. | |
| 4 | NUCLEAR SECURITY |
| 4.1 | DEFINITION: |
| Nuclear security involves safeguarding nuclear material and facilities from malicious acts, including theft, sabotage, and unauthorized access. | |
| 4.2 | AI’S ROLE: |
| 4.2.1 | Anomaly & Threat Detection: |
| 4.2.1.1 | ML models process multi-sensor feeds to catch patterns associated with intrusion, misuse of material, or cyber-physical attacks; and |
| 4.2.2 | Radiation & Material Sensors: |
| 4.2.2.1 | AI supports identification of radiation signatures and material type from complex detection platforms, improving response times. |
| Challenges for Security: Ensuring cybersecurity and resilience of AI/ML systems themselves; adversarial attacks or compromised data could undermine detection performance; and Ethical and privacy concerns when integrating AI with security surveillance systems. IAEA Perspective: The report notes ongoing work with Member States to develop frameworks and knowledge exchanges focused on responsible AI deployment in nuclear security applications. | |
| 5 | SAFEGUARD VERIFICATION |
| 5.1 | PURPOSE |
| Safeguards are the verification measures used by the IAEA to ensure nuclear material is used for peaceful purposes only under international agreements. | |
| 5.2 | AI APPLICATIONS: |
| 5.2.1 | LARGE DATA SCREENING: |
| 5.2.1.1 | AI/ML analyzes vast datasets — e.g., satellite imagery, environmental samples, radiation spectra — for patterns or outliers indicating undeclared activities. |
| 5.2.2 | AUTOMATING ROUTINE TASKS: |
| 5.2.2.1 | ML reduces inspector workload by automating repetitive analysis of surveillance footage or sensor datasets, allowing human experts to focus on anomalies of interest. |
| 5.3 | REGULATORY & ETHICAL ISSUES: |
| Use of AI in safeguards must align with privacy, legal, and data protection frameworks in Member States, which vary widely — a topic of ongoing regulatory analysis and planning. | |
| Outcomes Anticipated: Increased efficiency and depth of verification analysis; faster detection of discrepancies in declared nuclear activities; better allocation of inspector resources. | |
Perhaps the success factor for the collaboration is the core competencies of the AI and IAEA. Here are the core competencies:
- The intersection of AI and Nuclear Energy is already transforming the industry:
AI APPLICATIONS:
- Predictive maintenance & real-time monitoring;
- Fuel optimization & operational control;
- Cybersecurity and safety anomaly detection; and
- Digital twins & design simulation.
COLLABORATIONS:
- IAEA and NEA projects driving global coordination and trust;
- National labs like ORNL integrating AI with nuclear research; and
- Strategic partnerships aligning nuclear baseload power with AI energy needs.
WHY IT MATTERS:
- Improves safety, reliability, cost efficiency;
- Accelerates innovation & design; and
- Aligns with clean energy transitions and high computational demand from AI infrastructure.
