With the growing use of artificial intelligence by criminals, it's important to know how it's being used. A Cyber Security Intelligence post describes three ways.
Data poisoning is manipulation of a data set to make incorrect predictions. For example, spam could be labeled as safe content. Data poisoning could target an algorithm's availability or integrity.
Generative Adversarial Networks (GANs) are two AI systems pitted against each other to create content passing as original data. They can be used for password cracking, evading malware detection and fooling facial recognition.
Manipulating bots can trick algorithms into making wrong decisions.
The answer to offensive AI is to develop artificial intelligence defenses to deal with it.