AI in Cyber Security: Curse or Blessing?

Artificial intelligence, or AI, is one of the loudest buzzwords of our century. Back in the 1990s, the term was mostly associated with blockbuster movies about robots trying to take over the world. Today, AI is a very promising tool to help make people’s lives easier. Or much more difficult.

Almost no segment of our lives is free of AI. Retail, sales, and customer service are just a few examples of business uses where AI is already a major player.

Not to mention our personal lives: in a decade where even Home Alone gets a Google makeover, you shouldn’t be surprised when your apartment turns up the air conditioning and starts your coffee five minutes before you step through the front door.

No wonder cyber security gets its fair share of AI. But is it as good as it sounds?

Three Main Uses of AI in Cyber Security

In general terms, AI systems can be trained to look for malware or other suspicious activities. One way of training is through machine learning (ML), where the system is shown massive amounts of code, both harmful and clean. This way, the system learns how to distinguish between the two and how to react when encountering either.

Prevention

AI can be used for preventing cyber attacks.

Thanks to machine learning, AI is able to detect harmful code and separate it from safe algorithms.

In addition, AI can enhance protection by regularly checking the system for any vulnerabilities. If it finds anything out of the ordinary, it can alert an operator, or even fix it by itself.

Detection

Based on their preventive skills and steps, AI systems can detect any malicious or abnormal activity. With the power of machine learning, an AI system can teach itself what kind of red flags to look for. This function is especially important for systems that use a large number of connected devices, like a smart home that features many potential entry points for hackers.

Reaction

In case of a successful cyber attack, AI can enhance responsive measures as well. AI can check the system for signs of compromise, saving cyber security personnel a lot of time and manual labor. In addition, AI is able to isolate infected parts of the system or quarantine vulnerable data in a safe spot to prevent the attackers from accessing it.

The Drawbacks

While that certainly looks promising, there are side-effects of AI in cyber security.

Protective AI Compromised

We may think of AI as an infallible system. But just like any other system, AI is designed by humans and thus prone to error.

A system can be compromised right from the start. A human error in the code can lead to unforeseeable consequences later on. This can happen even without malicious intent. But if hackers get to the system during training, they can introduce their malware as a clean program, compromising the entire process from then on.

An additional danger lies in the very nature of AI: working without human intervention. An error can go undetected for a long time, allowing hackers access to the entire system without anybody noticing.

A way of defense against this is to use several, separate AI systems alongside each other, each one constantly checking how the other is working. While this has definite advantages, there’s an undisputed drawback: the cost.

The math is obvious: the more AI systems in use, the higher the cost.

Hackers Using AI

Unfortunately, any tool used for good causes can be used for bad ones just as well. AI controlled by hackers can learn the way anti-malware software (or even AI) detects suspicious activity and potential cyber attacks. Then, using its knowledge, it can alter its malware’s appearance or code to get past the defenses.

Even more alarming is the fact that hackers may infiltrate an AI system and change the way it works. For example, if a factory’s AI that runs its machinery is infiltrated by malicious intent, the consequences may be dire. A simple malfunction can lead to grave injury and eventual factory shutdown within a short amount of time.

With a coordinated AI cyber attack, whole industries and even national economies may be brought to their knees.

Does this sound like the plot of a Hollywood blockbuster? Certainly. But unfortunately, life may start imitating art in this regard very soon.

We don’t want to sound overly pessimistic. AI is a great opportunity and a field that’s certainly worth exploring for many businesses. However, today it’s more important than ever to employ sufficient cyber security measures so you can keep all your systems safe.