WASHINGTON — A top goal this year for the National Security Agency’s nearly two year old cybersecurity directorate is to scale up and expand its partnership with the defense industrial base.
After a series of high profile hacks and breaches, the NSA had to “up its game,” officials said. It established the directorate as a means of using its unique intelligence capabilities to share threat information with companies and the defense industrial base in a timely fashion to ensure they stay ahead of the most sophisticated threats.
“Since we established the cybersecurity directorate a little over two years ago, we’ve been doing great things. We’ve been using intelligence and letting it drive risk mitigations and building up partnerships,” Neal Ziring, the directorate’s technical director said during a virtual presentation Feb. 1 hosted by INSA. “How do we do that at greater scale? How do we expand what we’ve been able to do with the defense industrial base, for example, from 100 companies to 1,000 or even 10,000? How do we start getting cyber threat information out to lots and lots of partners and doing it at relevant speed?”
In times of heightened international tensions, such as the current imbroglio with Russia, Ziring said cyber threats can increase.
Working with other federal partners such as the Department of Homeland Security’s Cybersecurity and Infrastructure Security Agency and the FBI, Ziring said his organization has used its unique intelligence insights to provide a fuller picture of what adversaries are doing and help defeat them.
“One of the things we’ve done in that regard — directly as a result of these heightened tensions — is issuing additional guidance published on nsa.gov. We’re published jointly with CISA, FBI and others about specific threats and tradecraft of Russian actors and how to secure some of the systems that they tend to go after,” he said.
Securing artificial intelligence and machine learning
As more Department of Defense entities are looking to incorporate machine learning and artificial intelligence into their practices and systems, NSA is in the beginning stages of understanding the security needs required to do so. NSA will be digging deeper under the hood to understand the security vulnerabilities across the entire lifecycle for applying machine learning or artificial intelligence into a system — such as a driverless vehicle — Ziring said, citing an example of how DoD might employ such technologies.
Currently, it’s not clear if traditional security measures with normal software would be enough to secure these technologies.
“An AI or ML system can be so dependent on the things you do early in the lifecycle, like gathering training data and training initial models. We have to extend our security thinking back into those stages,” Ziring said. “That will be a focus for us in the next few years, is thinking about how we can do that, do it better and then advise our customers as they adopt that technology, how to do it secure.”
Mark Pomerleau is a reporter for C4ISRNET, covering information warfare and cyberspace.