To elaborate on leakage: it's a case where something in the training or test dataset wouldn't be available in a production setting.

As a funny example: I remember we were once given a dataset to predict prostate cancer from ~300 variables. One of the variables was "had prostate cancer surgery". Turned out that was a very good predictor of prostate cancer ;). Thankfully that was an example where we caught the leakage. Unfortunately there are cases where we don't catch the leakage.