AI Nuclear Waste Sorting System NucSort Deployed in France: 99.7% Accuracy in Classifying High-Level From Low-Level Waste

French nuclear company Orano announced on May 30, 2029 the successful deployment of AI nuclear waste sorting system NucSort at its La Hague nuclear fuel processing plant. Combining neutron imaging technology with deep learning algorithms, the system performs precise radioactivity analysis and classification of each piece of nuclear waste with 99.7% accuracy.

Accurate nuclear waste classification is fundamental to nuclear safety management. High-level waste requires deep geological disposal, while medium- and low-level waste can undergo near-surface disposal. Misclassification can either create safety hazards or result in unnecessary disposal costs.

NucSort reduces per-item sorting time from 45 minutes (manual operation) to 3 minutes while improving accuracy from 85% to 99.7%. Orano estimates that at La Hague alone, NucSort will reduce approximately €20 million in excess disposal costs annually. The system has attracted interest from the UK Nuclear Decommissioning Authority and Japan Atomic Power Company.