Artificial Intelligence and Extended Reality May Pose Security Risks, Expert Warns

Payton predicted that "AI poisoning" will be something to be concerned about in 2021. As Towards Data Science notes, a "poisoning attack happens when the adversary is able to inject bad data into your model's training pool, and hence get it to learn something it shouldn't." In solidly built AI models, Payton noted, "your [AI] coach should be self-learning and contextually aware and almost become a black box to the engineer" once it gets up and running. "My prediction is that, as we're implementing more AI, hackers will hack in and change that algorithm undetected, so that the AI will do things not initially in the design," she said. "AI is going to be cybercriminals' weapon of choice, to help them crack into more accounts, networks and data stores."