Researchers think that adversarial examples could help us maintain privacy from machine learning systems

Open in new window