About IAEA
The International Atomic Energy Agency is the world's central intergovernmental forum for scientific and technical co-operation in the nuclear field. It works for the safe, secure and peaceful uses of nuclear science and technology, contributing to international peace and security and the United Nations' Sustainable Development Goals.Activities by IAEA
Machine learning for detector signal discrimination
GlobalThis project assesses how machine learning on FPGA devices can help discriminate neutrons from gammas in a detector. The IAEA contributed measurements for the study, which aims to improve quantitative results and enable real-time discrimination.
Activity Type Research/Reports/AssessmentsAI Tools/SolutionsAI and Machine Learning to Support Capacity Building in Nuclear, Radiation, Transport, Waste and Emergency Preparedness and Response
Global Not Applicable
This project focuses on the use of AI to support IAEA nuclear safety, emergency preparedness and response, and regulatory capacity building activities, including peer review and advisory services. These activities require the analysis of large volumes of heterogeneous information, such as national self assessments, emergency arrangements, regulatory frameworks, technical reports and mission findings. The project examines tailored AI enabled tools to assist with document analysis, knowledge retrieval, consistency checking and synthesis of expert inputs, with the objective of improving efficiency, traceability and knowledge retention while preserving expert judgement.
One tool, called Robotox, is for e-learning courses: it is an agentic AI prototype assistant which can adapt to a learner’s level (from beginner to advanced), deliver scenarios, and generate quizzes. In the context of radiation protection, the project engages international experts to share knowledge and best practices on AI-radiation protection in medicine among Member States. One sub-project focuses on the use of AI for public communication in nuclear and radiological emergencies. Overall, the work emphasizes domain specific, human in the loop applications designed to strengthen Member State capacity, support timely mission outputs and enhance the effectiveness of IAEA support. The project is currently focusing on testing and scaling these tools to benefit Agency staff and ultimately Member States.
AI-assisted review of satellite imagery and site maps
Global Not ApplicableUse of Machine Learning to assist in the detection and characterization of IAEA nuclear safeguards-relevant infrastructure in satellite imagery, reducing manual effort and prompting the analyst to focus on areas of potential change. Use of modern multimodal models and segmentation techniques to recognize buildings and attributes in unstructured site maps and convert them into a vectorized, georeferenced format suitable for further analysis.
Activity Type AI Tools/SolutionsMachine learning for safeguards surveillance
Global Not ApplicableUse of AI to assist inspectors during the review of images and data from safeguards surveillance and monitoring systems.
Activity Type AI Tools/Solutions