Throughout the past two years, we have seen breach after breach. From the Yahoo breaches to the international cyber-hack involving the NHS in May 2017, they never seem to stop!
So what can be done about it? This year, we expect to see the rise of machine learning to identify attacks quickly. Unlike the ransomware that crippled our NHS along with other organisations in 98 countries across the globe on the 12th May 2017, not all data breaches flash on your screen with a frightening message demanding payment next to an intimidating timer. Many data breaches occur without companies being ever aware of it. Like the Yahoo data breaches, it took them years to realise their security systems had been compromised.
Hidden criminals
The cyber-criminals are often portrayed as young men in hoodies hiding in a dark room, and they don’t want to be found. Their anonymity is crucial and their work is best done in secret. Illegally stealing information or spying on data without triggering security tripwires means that the cyber-criminals can sell or misuse the information they obtain more freely. Once a company finds out and clamps down on security, data owners will usually be warned by companies and on high alert for suspicious activity.
Nowadays, these security tripwires aren’t as simple as they seem. Cyber-criminals can hide amongst the huge number of genuine user interactions so that security systems may not identify anything suspicious. However, no matter how genuine they may seem, there will often be a ‘tell-tale’ usually in the form of patterns. Unsurprisingly, these patterns are incredibly hard to find; perhaps impossible for a human to distinguish. However, machines can be programmed to learn how to find these patterns.
Machines don’t make room for human error
Machines can go through a tonne of information within minutes, unlike humans who can take years to go through information with a fine toothed comb. Security professionals and experts have, over the years, developed all kinds of network intrusion systems to look through data in search of these patterns. But this isn’t enough. With software programming becoming more and more complex, security experts can’t keep up; but what they can do is program the machines to learn to improve its ability to recognise patterns.
Cyber experts can train a machine into learning certain algorithms to identify signs of attacks. At the initial stage, experts can start with the most simple: brute-force attacks. These attacks work by using software to insert a vast number of combinations to find a key or a password.
This is why we tell you not to use a simple password like ‘guest’.
The machine can be programmed to recognise all sorts of known brute-force attacks as well as genuine data. Over time, the machine can learn to identify which is which. Security experts hope Machine Learning will continue to develop and progress to a point where it will be available on the market for companies for protecting their data.
Drawbacks
However, these learning machines aren’t going to simply solve the problem of cyber attacks. Let’s not forget that these cyber attackers won’t go away. When companies have access to these machines and its advanced programming, so do cyber-criminals. No matter what kind of advanced lock you use to secure valuables, criminals will always find a way around it. However, that doesn’t mean to say we should give up and let cyber-criminals take over.