Overfitting is a common challenge in machine learning, where a model learns the training data too precisely — including noise and outliers — and fails to perform well on new data.