The Pentagon's Joint Artificial Intelligence Center explores ways
that adversaries can exploit AI vulnerabilities
By infiltrating the training data collection and labeling process, enemies can "poison" the model and exploit its consequent weakness
Click here for related article