On February 10, 2026, a seminar titled “Trustworthy AI for the Grid: Verification-Informed Neural Networks for the ACOPF Problem” was delivered by Dr. Maad Alowaifeer (Assistant Professor, Electrical Engineering Department, KFUPM | Acting Director, SDAIA–KFUPM JRC for Artificial Intelligence) at the Electrical Engineering Department, KFUPM.

During the seminar, Dr. Alowaifeer discussed the growing need for AI solutions that can support power-system operation without compromising safety. He explained that many learning-based approaches may produce fast results, but can still violate critical operational constraints—an issue that limits their use in real-time grid applications.

Dr. Alowaifeer then presented a verification-informed neural-network approach and described how verification can be incorporated into the training process to reduce worst-case violations. He also highlighted how the proposed framework leverages mathematical relaxations to enforce constraints more reliably, and shared experimental observations showing improved feasibility compared to standard black-box models.

The following are the highlights of the event: