AI needed to vet 100 billion cyber threat items per day

‘Minority report’ level tech may soon guess cyber crimes before they happen

An information analyst works in front of a screen showing a near real-time map tracking cyber threats; California, December 29, 2014 (photo credit: REUTERS/BECK DIEFENBACH)
An information analyst works in front of a screen showing a near real-time map tracking cyber threats; California, December 29, 2014
Artificial intelligence is an absolute imperative to vet obscene amounts of data and cyber-threat intelligence, up to 100 billion items per day, according to an AI expert at a conference analyzing how the issue impacts national security.
Derek Manky, chief of security insights for global threat alliances at Fortinet, made his remarks at a virtual conference co-sponsored by the Institute for the Research of the Methodology of Intelligence and Israel Defense.
His company has five million devices constantly collecting cyber-threat intelligence data around the world, while working with NATO, the International Criminal Police Organization (INTERPOL) and other major national and global security agencies, he said.
They also use new AI technologies and strategies, including computer “Game Boards” and “heat maps,” to chart out ongoing cyberattacks so that they can advise clients how grave they are, Manky said.
Fortinet is able to do this by using AI to chart out specific hackers’ tactics and typical time frames for collecting and invading less-important systems before they try to penetrate into more critical systems.
Targeted networks would prefer to never be penetrated, and using this AI-based threat evaluator allows them to divert their key resources to more serious penetrations and invest fewer resources in combating lower-grade cyber intrusions.
Just over the horizon, by analyzing attackers’ patterns, Fortinet’s AI models should be able to predict the time frame of attacks once started and even guess the identity of likely victims before they take place, Manky said.
The conference also included former top IDF Intelligence officials, such as conference chairman Brig.-Gen. (ret.) Yossi Kuperwasser.
Former IDF Intelligence Col. (ret.) Itai Shapira explained how new technologies such as AI were both killing the old way of performing intelligence gathering and analysis, while providing a new basis for reinventing the medium.
“AI is not a new technology for one part [of the intelligence process],” he said, “but something that advances connections between all the different intelligence disciplines.”
AI would create opportunities to leap forward in efficiencies in gathering intelligence and reframe how and when different intelligence processes and decisions should take place, Shapira said.
Within about a decade, the entire intelligence arena would be transformed, he predicted.
Dr. Ehud Eran, a senior fellow at the Israel Intelligence Heritage and Commemoration Center (IICC) and a professor at the University of Haifa, said AI’s real potential still has not been realized.
To date, most AI achievements have been at the tactical intelligence level, he said.
In the future, AI could change the way intelligence communities and countries view and address the world strategically, Eran predicted.
He recommended a more radical disruption of the traditional intelligence process to rebuild it even better with AI.
Dr. Shay Hershkovitz, a senior fellow at IICC, said another important lesson already learned from developments in AI would be a change in who would work in the intelligence community in the future.
Future recruits for the intelligence community needed to be as tech-savvy as they were capable in becoming experts in analyzing foreign national-security issues, he said.
This would be a major departure from past eras in which tech-savvy recruits were limited to small numbers in a practically segregated and separated small technology department.