- Researchers from the Massachusetts Institute of Technology have created an AI system that can predict a cyberattack before it happens in 85% of incidents.
- MIT’s Computer Science and Artificial Intelligence Lab built AI2, which employs three different machine learning algorithms to detect suspicious events.
- AI2 can review data from "tens of millions" of log lines each day and flag suspicious incidents, researchers said. Then, a human takes over to check for signs of a breach.
MIT’s work with AI2 was presented last week at the IEEE International Conference on Big Data Security. While it’s not the first time AI has been used to weed through such data to try to predict cyberattacks, AI2 appears to be doing so more successfully than its predecessors.
AI2 can’t do it all on its own—like most AI programs, it can weed out "suspicious" elements, but it needs human interaction to determine whether or not those events are actually suspicious. But the program also refines its internal models based on the human reaction, so it eventually becomes better able to differentiate benign threats from real threats.
"The more attacks the system detects, the more analyst feedback it receives, which, in turn, improves the accuracy of future predictions," saidMIT research lead Kalyan Veeramachaneni. "That human-machine interaction creates a beautiful, cascading effect."
Given the rapidly expanding threat landscape, a system like AI2 could help companies identify and react to cyberthreats more quickly and more easily determine if they are dealing with a potentially serious attack.