Adversarial Examples: inputs that have been specifically designed to fool machine learning models into making incorrect predictions.