How Machine Learning Affects Digital Security

Elliot Thompson

You might think the computer virus is a problem of the last couple of decades, but in fact, the very first virus was devised back in 1982. It was intended primarily as a prank, and little did the inventor of Elk Cloner know what he was starting and how far computer viruses would advance in the future.

At the time, the medium of transmission was infected floppy disks, but once the World Wide Web took off, so did the spread of viruses. Very quickly, in response to this, anti-virus software began to appear, led by pioneers including John McAfee and Eugene Kaspersky, names that need no introduction today.

But however much ingenuity goes into combating hackers whose sole aim seems to be to cause as much disruption as possible, the hackers have always stayed at least one step ahead. One need only look at the plethora of serious cyberattacks occurring today to see this.

In light of this tidal wave of menaces, many experts in the cybersecurity industry and beyond are pinning their hopes on machine learning to provide at least a partial solution. The self-developing nature of machine learning, compared to the more man-made basis of artificial intelligence, does suggest that it would be a good way to filter the unimaginably large amounts of data that are being constantly generated today.

It’s also a technology that is being used effectively today by businesses like Facebook and global bank HSBC for a variety of purposes, including service personalization and customer identification.

The theory is that because machine learning is a very efficient way of spotting patterns and more importantly, identifying when those patterns are disrupted, it could be just the sort of early-warning system needed to isolate attacks before they take hold. But one caveat is that there is so much activity occurring at any one time that the number of “false positives”—suspected risks that turn out to be safe after all—could bring many systems to a halt.

One solution that has been put forward is that machine learning should be used to flag these warning signs, which can then be investigated by a human security analyst to assess whether they present a genuine threat.

While this would undoubtedly help, preventing viruses and other attacks from getting on the system in the first place is obviously important too. That’s why organizations are focusing more than ever on making staff aware of best security practices, with many also using VPNs for the greater security and anonymity that they offer.

So while cyber threats are certainly never going to disappear completely, there is hope that a combination of cutting-edge technology and human contributions may block many more before they can do serious damage.

For more insight on cybersecurity, see The Evolving Role Of Security In Today’s Ever-Connected World.


Elliot Thompson

About Elliot Thompson

Elliot Thompson is a digital security expert. While staying on top of clients' information technology needs, he also writes about the subject to make people more aware of any potential risks.