The Impact Of Artificial Intelligence On Cyber Security [+Examples]

The Impact Of Artificial Intelligence On Cyber Security [+Examples]

It’s one cool evening.

You are from work and for decades you’ve been planning to hit the gym.

Today seems like the perfect day. Except you don’t know where any gym is. So you pull out your smartphone, launch Google Maps and search ‘gym’.

In no time, Google has a couple of options of nearby gyms. But not just that. You tap on the first one on the list and now it shows you various routes, telling you the shortest route based on whether you are going on foot, private transport or public transport.

Ready to launch your cyber security career with Udemy? Get started TODAY for just $9.99 (95% OFF) with my link below:

How’s this even possible?

How could Google know about your local town to such detail? Did the CEO visit someday and order a sweep of the roads and gyms?

Well, that’s artificial intelligence in action.

Artificial intelligence has permeated almost all areas of our lives today, significantly making life much easier. In fact, here are other sectors where AI is trending in 2021.

In this article, we are going to look at how artificial intelligence has impacted cyber security.

And while at it, we’ll discuss the ways in which AI has been applied in cyber security to make more secure systems, as well as the ways through which AI has made securing applications even harder.

I hope that by the end of this article on the impact of artificial intelligence on cyber security, you’ll have a clearer understanding of the role AI plays in cyber security in 2021.

Artificial intelligence simply refers to the ability of machines to imitate human-like characteristics like learning and making decisions. However, that’s a rather simplistic definition. If you want to under it better check out my article where I define AI (+how to get started) in more details.

So let’s get started.

Benefits of AI In Cyber Security

In this day and age, AI technologies are widely being used to confront some of the most common cyber security threats with a high degree of success.

By using applications that can learn from their environment, a lot of threat detection and thwarting have been automated.

(Cough)

Hackers can use the same technologies to automate their attacks.

First, let’s look at how artificial intelligence has positively impacted cybersecurity.

1. Threat Detection

The typical job of a cyber security engineer goes like this…

Detect an attack, thwart it and get things back to normal without necessarily affecting the day to day operations of the business. Basically, the marketing team upstairs shouldn’t even notice that something happened (pun intended).

That’s overly simplified but you get the point.

So how has AI impacted the ability to detect threats before damage is done?

Cybersecurity engineers have used machine learning, a branch of artificial intelligence, to identify, analyze data and detect threats early on, before they can exploit vulnerabilities in the system. 

Machine learning gives computers this human-like intelligence where the applications are able to learn and adapt themselves based on examples, the data it receives and past experiences.

Loaded with this superpower, the machine can then predict threats by simply analyzing anomalies. And this it does with a super high degree of precision.

And because it’s a machine, it doesn’t need to take a coffee break and have small talk with colleagues. So it’s on the watch 24/7. This will definitely beat a team of 100 security engineers because, well because engineers are human. 

So even if there was a sudden surge in threat volume, it still gets detected.

As you can see, AI has positively impacted cyber security when it comes to threat detection. 

It squarely addresses the reactive nature of cyber security where engineers sit back, sip black coffee and wait for an attack to happen so that they can react to it… 🙂

2. Authentication

What is your Facebook password?

No really. Tell me. Haven’t you been using the same password for years and years on end?

Heck, it could even be the same password that you also use for your internet backing account, your Gmail and everything in between.

I get it.

I am in the same boat too. We are consistently told to use strong passwords. Scratch that. I consistently tell my colleagues to use strong passwords with mixed numbers and letters, with caps and special characters.

But who can really remember such a password? It means that you then have to really on password management tools to bring your passwords along with you. It’s the reason passwords are an easy attack point as attackers can make a couple of guesses and they are in.

One of the major advances of cyber security when it comes to authentication and password protection is to use face recognition to log you into your devices. 

At the onset, face recognition was a hit or miss as a hairstyle change could mean it’s not you and your device would deny you access.

Well, not anymore.

Software engineers are now using artificial intelligence to enhance facial recognition systems and make them reliable. Unlocking your device with your face doesn’t feel like pulling teeth anymore.

A great example is the use of face detection on Apple’s iPhone X devices called Face ID.

Through machine learning and neural networks, Face ID enhances biometric authentication by processing your facial features using infrared sensors and building a model based on key correlations and patterns.

One main advantage is that this artificial intelligence-based software also works under different lighting conditions and can compensate for inevitable changes like:

  • growing facial hair, 
  • changing your hairstyle or even 
  • wearing a hat.

In fact, Apple claims that with this artificial intelligence technology, there is only one in a million chances that you could open your device with a different face.

So clearly AI has come to the rescue of password thefts and related cyber security attacks.

3. Behavioural Analytics

You don’t use Facebook that much.

Often you log in for a few minutes after 5 PM, scroll through your feed, like one or two photos and you are out, till the next day.

Then one day, there is this you that logged in at 5 AM in the morning, is aggressively sending friend requests and messaging contacts through messenger. Then from nowhere, Facebook logs you out and sends you an email that there was unusual activity detected on your account. 

So your account was temporarily disabled for your own security.

Talk of artificial intelligence at work in stopping hackers right in their tracks before they cause significant damage.

Behavioural analytics is another significant enhancement to cyber security that is used to nip common network security threats in the bud, thanks to artificial intelligence and machine learning.

How does it achieve this?

Using machine learning algorithms, a computer can observe, learn and then create a predictable pattern of your behaviour based on how you use your devices or online platforms.

And like we already saw at the beginning, these behaviours could range from your login times, scrolling patterns and even your IP address.

The AI algorithm then sits down, leans back, fingers crossed and keeps guard… watching your account with a hawk eye.

If it detects any behaviour that falls outside the patterns it has established through unsupervised learning, it pulls the plug.

An activity that could rattle the ML algorithm could be something as simple as a sudden increase in your typing speed, and actions could range from:

  • temporarily suspending your account, 
  • flagging it for a human review or 
  • permanently disabling it.

Imagine a situation where this could be useful.

There has been a data breach, followed by massive password theft of your sensitive accounts.

This data is sold on the dark web. An attacker gets hold of, say the Facebook password, logs in and tries to scam your friends or followers. This AI algorithm that has been trained through machine learning is able to put a stop to this before it becomes anything.

So, as you can see, artificial intelligence comes to the rescue again.

Challenges of AI In Cyber Security

The advancements that artificial intelligence has made in cyber security are by far and large genuine and significant.

Having mentioned that, there is a high probability that black hat hackers could weaponize AI technologies like machine learning and deep learning, thereby using them to advance their attacks to a wider and worrying extent.

You might now feel lost thanks to these terms I keep throwing around. If you want clarity, why not check this AI vs ML vs DL comparison to know the difference between these technologies?

In fact, these three terms have been trending steadily for the last five years, according to the Google trends screenshot above.

Apart from that, there is still a huge gap between the theory of AI in cybersecurity and it’s practicality for small business.

Is it a reserve for the big tech?

Let’s look at some of the threats and challenges that AI poses to cyber security.

4. High Operational Costs

Implementing an AI-powered cyber security solution is damn expensive.

No really.

That’s why while it’s a priority for the big tech companies who are already raking in billions of dollars in profit each year, it is almost like an afterthought for small to mid-size businesses.

Let me explain how.

To begin with, finding highly skilled artificial intelligence engineers with experience in cyber security is super hard. And this is because AI is a very highly skilled profession, often requiring a lot of time and patience in study, research, experiments and just being smart overall.

You can bet not many people are going down that road, just to earn a living. In addition to that, when you find them, you pay an arm and a leg in salaries. Artificial intelligence is rated one of the highest pay careers in tech today. With 47% of SMBs saying high costs is stopping them from embracing it.

As if that is not enough…

Implementing operational machine learning models that can actually yield real results requires an extremely high amount of computer processing power. So a huge initial investment has to be made in establishing a data centre with super-fast computers.

Oh, and there’s one more thing.

Machine learning models need a huge amount of dataset to feed on so that you can self-train, learn and be able to predict and make decisions. Where does a startup get such an amount of data from? 

So as you can see, while AI has presented us with an amazing approach to curbing cybercrime and threats, it has also raised the bar too high, drawing a line on the sand as to how can implement it and who can’t.

If you can’t implement it, you are left vulnerable. Your only option is to embrace a knee jerk reaction whenever an attack happens.

Check the next challenge to even get a better idea of what I’m talking about.

5. Mutating Malware

One of the reasons AI has advanced so fast is because of its open-source approach.

Artificial intelligence tools, libraries and models are all free and open source. Anyone with a computer and an internet connection can download these tools and start using them.

First, this is a good thing because the open-source community has developed some amazing tools to further advancements in this sector.

So what’s the problem?

(Drum roll)

Malicious hackers can download these libraries too.

They are not a monopoly for well-intentioned users. Worse still, these malicious hackers could actually be highly skilled AI engineers and cyber security experts who have recently been fired from their jobs.

They know exactly how the system works and are best placed to exploit their vulnerabilities.

Let’s take an example of threat detection.

Using machine learning models, I can build a system to scan, detect or predict attack probabilities based on certain patterns. So here is what one would do. One would build his malware to be able to learn from the system it is installed in, imitate the behaviour of the user to avoid being detected by the native AI-powered threat detection.

The malware is able to mutate and adapt itself to imitate its host behaviour thought machine learning. In the meantime, it is sending sensitive data to a remote server, in preparation for a major attack once enough information is collected.

Talk about setting a thief to catch a thief.

It is one of the clears ways that AI has impacted cyber security in a negative way.

Conclusion

It is clear that AI has played a huge role in making cyber security attack preventions much easier and predictable.

And this is an amazing thing for cyber security professionals and penetration testers like me, who have to deal with uncertainty, risks, threats and the fear of them every day.

However, ordinary users are not as enthused by this. 

A good number of citizens are concerned about the growing trends in automation and whether these will finally take the place of human decision making.

But the good side of AI applications just makes it unstoppable right now.

It is a very lucrative career and critical thinkers, analytics thinkers and problem solvers are jumping on board to lend a hand in developing sustainable technologies.

If you want to get started with AI and see what you can come up with, I have compiled some of the best FREE resources for learning AI in 2021 to get you started.

These resources will not only provide the theory that you need, but you’ll also get a lot of hands-on practice projects to cement your learning.

Who knows? May at the end you’d be able to provide a solution to some of the threats sloppy AI applications have brought into cyber security.

Have you used any AI application before?

What are your thoughts on the place of artificial intelligence in modern society with respect to cyber security?

Please share your thoughts in the comments below.

Lerma Gray

Hey, I’m Lerma, a data analyst with experience in intelligence tools like Power BI. On this blog, I write about my experience with the various techniques that I interact with on a daily basics. This ranges from software development tools, cybersecurity best practices and artificial intelligence. Let's connect in the comments below.

Leave a Reply