Scientists Develop New Machine Learning Models to Assess Uranium Enrichment Levels

According to a Phys.org report published on November 10, 2026, scientists have developed a new AI system capable of predicting uranium enrichment levels with extraordinary precision while identifying different types of nuclear material using gamma spectroscopy data. The system was built on tests of three machine-learning models: XGBoost, CatBoost, and LightGBM, whose combined predictions the researchers said yielded accurate and reliable results for the task. The researchers indicated that the findings could offer the U.N. nuclear watchdog valuable new tools to strengthen its safeguards under the Nuclear Non-Proliferation Treaty (NPT), helping inspectors assess declared nuclear material more effectively. No details were provided on the dataset used for training, the models' error rates, or comparisons with existing analytical methods, and no timeline was given for operational deployment by any inspection body. As of this report, the research has been presented only as a scientific result, and there has been no official comment from the relevant international monitoring organization on adopting the method.