A newly discovered set of critical vulnerabilities in a machine learning framework known as TorchServe could allow cyberattackers a way to completely subvert artificial intelligence (AI) models for a ...