Remote Code Execution

nltk is vulnerable to Remote Code Execution. The vulnerability is due to models containing pickled Python code, which could allow an attacker to execute arbitrary code. An attacker would need to preform a man-in-the-middle attack to modify the packaged pickles such as the averaged_perceptron_tagger and punkt models, or trick a user into downloading an arbitrary model.