This paper introduces a novel method for design validation that merges Fault Simulation with Artificial Intelligence, leveraging Machine Learning techniques. Initially, we used all available Test Vectors to verify the Design Under Test (DUT) by applying standard Fault Simulation methods such as Stuck@0 and Stuck@1, which serve as the training patterns for the ML model. These vectors, processed through an EDA Simulation tool, simulate stuck faults on each input signal and log the outputs. This logged data then feeds into an AI/ML framework to develop a predictive model. The model is trained on a randomly selected 20% subset of the data, using densely layered neural networks optimised with activation functions like ReLU and Sigmoid, and fine-tuned through hyperparameters such as epoch length, accuracy, data loss, and learning rate. After training, the model is tested for robustness against the remaining 80% of the test vectors, using heat maps and other graphical tools to assess performance and ensure result validity. Subsequent validations of the model account for any changes or updates to the DUT, with tests conducted using only 20% of the updated test vectors to predict outcomes on the larger 80% portion. Comparisons of these results with initial runs highlight any discrepancies and fault statuses effectively. The methodology proposed not only significantly cuts down on the duration of simulations but also reduces reliance on extensive testability tools, enhancing efficiency in the validation process.