Artificial Intelligence in Cyber SecurityUpdated: November 02, 2019
Data is at the core of the Internet, hence it's also at the core of cyber security. Valuable resources (time and human work force) would be put into analyzing all the amounts of data, so why not use something that can save you both?
This is where Artificial Intelligence comes into place (on of the domains, of course). I've heard at an Adobe webinar that AI should be used to help people get rid of the mundane, repetitive tasks, in order to focus more on innovating and creating. I truly agree; it's one thing to get yourself buried in files, with your eyes red from concentrating on them, and other to use AI to analyse those files, and you only to come up with scenarios, tests, and final results.
And what other better place for AI to "spread its wings" than cyber security? It already started for a couple of years, but in 2019 it's still in the beginning, as the whole AI industry is still being tested and tried in more and more domains.
In this article you'll learn about what is artificial intelligence, what is cyber security, and how can they both work together.
When people say "Artificial Intelligence" I'm sure they are thinking about robots, and even robots taking over the world. Still, we aren't quite in that place of time, nor we'll be (talking about robots taking over the world). There are robots that are quite human-like, such as Sofia, but that's about it.
Artificial Intelligence refers to the virtual capabilities of a machine to act like a human: learn, reason and self-correct. Of course, an AI system is created by humans; they write code, input all the data they need, and create processes that will be able to simulate the human intelligence. These machines don't get tired, don't have needs, and they certainly are more capable of doing things fast and more efficient than humans.
We began hearing about AI mostly in movies, the science-fiction/fantasy ones, from more than 5 years ago: Transcendence, Chappie, Ex Machina etc. Most of the focus either on using AI to create self surviving machine that can, at some point, do harm to humans, either on robots, either on bad robots.
The reality, though is a bit different. Companies began to use Artificial Intelligence for those time-consuming processes that I mentioned above. Artificial Intelligence has been used (in various forms) since the '70-'80s. From then, it only evolved. Apple is using it for Siri, Microsoft for Cortana, and so on. But still, don't think that these are the only areas; Google just rolled out a new update for its algorithm, that basically uses Artificial Intelligence to scan not only the key, main words, but the whole query and the relation between the words in it. Adobe is using Artificial Intelligence for Adobe Sensei, a powerful machine that brings deep insights, enhanced creative expression, accelerated tasks and workflows, and real-time decisions (source: https://www.adobe.com/sensei.html).
Artificial Intelligence can be used for the following:
- news feed experiences, tailored for the user (Facebook, Twitter, Instagram etc.)
- product ordering, and refining them constantly for a maximum engagement;
- predictable and repetitive tasks;
- and many more.
Because it's a "method" that can be used in so many domains, Artificial Intelligence can be broken down into three main categories:
ASI - Artificial Super Intelligence is the most powerful, and can be that one which "scares" people. The human brain is limited to its "collection" of neurons, and to what it learned throughout the life, what it experienced and so on. It's already known that a person is only using a small part of his brain's capacity, even thought some other says that's a myth. Either way, we are still not powerful enough to think of situations, predictions, abstractions at a super high level; we don't have the resources. But a machine, powered by ASI, can; the machine it's not limited to basically anything, only to a power source, maybe not even that. This leads to calculations and tasks far superior than what a human can do, leading to a computer's cognitive abilities superior than the human's.
AGI - Artificial General Intelligence is the type of AI that can be applied to multiple things. It has the ability to learn from previous tasks, and improve itself. AGI is the one depicted in movies, such as Her. For now, in real life, it has been used to play the game Go, and it has become the best player in the whole history; it was the first to defeat a Go world champion.
ANI - Artificial Narrow Intelligence is the one that you're encountering on a day to day basis, as you have it on your phone. As the name suggests, it only focuses on one task, but it has to perform it with intelligence. "Hey, Siri!" sounds familiar? Well, ANI is being used for speech recognition, voice assistants, crawling web pages etc. Devices using ANI perform the task in real-time, but they need a data-set to pull out the information for completing the task. We are surrounded by this type of ANI, as this is the type which is being used in basically everything.
We all know what security refers to, and we all know what cyber stands for. Just to have this article properly written, I'll point some details about each of the terms. Security means protection; it means to protect whatever you need from various types of attacks (PIN codes for your credit card, Touch ID for your iPhone, passwords for accounts, chains for gates and so on). Cyber refers to the cybernetic https://www.merriam-webster.com/dictionary/cybernetics, it's where the term comes from. It encompasses computers, technology, virtual reality etc.
Having these said, cyber security refers to protecting all the systems and the networks, along with programs, from digital attacks.
As we now live in a world that is "digitalised", we not only have to care for our physical possessions, but also for our digital ones. Of course, there are only digital attacks out there, there are also flaws in the systems, that can be discovered and used to cause harm. The harm that can be done is translated in stolen data, which can cause further damage: attackers ask for ransom, attackers are exploiting the stolen data, and so on.
Cyber Security is important for both the individuals and the companies. As individuals, we are mostly targeted for credit card data, through various attacks. Companies, on the other hand, are exposed to data breaches, for example, which can cause money loss.
As I mentioned in one of my articles - it's pinned below - most of the time, an cyber attacker will attack if he knows he has something good to gain out of it. Pretty similar to a classic attacker, right? Your social media accounts will not bring the cyber attacker any good, unless you're a celebrity; then he can ask for a ransom. Banks, army facilities, companies of all sizes are much more prone to be attacked, hence they will invest much more money into protection.
Here's another article that will help you better understand the importance of cyber security :
Artificial General Intelligence and Artificial Superior Intelligence are both some things for the future. For now, when we say AI in Cyber Security, we're talking about Artificial Narrow Intelligence.
As I've described above, Artificial Narrow Intelligence is capable of performing one task, using a data-set from which it pulls out information. Its advantage is seen in the fast way of processing millions of data, a thing that humans can't do.
Hackers are already using forms of AI, wanting an easier way to perform their harming tasks - for example bots that are specifically created to break passwords; as the human will be able to do this in a number of days, a well designed bot can accomplish the task in a matter of hours. It just needs to know exactly what it has to do, and the scenarios that can appear, after which it can continue the job alone, while the hacker only waits for the results.
And while hackers are using AI, why shouldn't companies use AI to protect themselves? Of course, the smaller companies don't necessarily invest in a AI security systems of their own, because it's not that cheap. Bigger companies, though, are putting up security systems that can catch any attempt of an attack. They have their own security programmers which do only this: create security systems, monitor them, repair them, test them and so on.
There are a number of ways in which AI can be used for a company's cyber security:
- biometric logins; passwords can be cracked, even the most powerful ones. Using biometric logins, such as fingerprints, retina scan or others, companies can make sure that their systems will have a higher level of security, without that high possibility of a hacker cracking a password.
- detection of malicious activity; it's better to prevent than to combat, right? AI used for this can scan through thousands, millions of actions, and point out the ones that seems suspicious. Some powerful systems can even detect malware before it actually gets to a computer, just by scanning the behavior. Also it would be easier and faster to find out what email contained a virus, for example, who sent it and so on; the IT team can do this without AI, but it would take a bit longer.
- Natural Learning Processes; which can help with always being informed about what's new in the cyber security world, helping the security team stay prepared. It's one thing to set up notifications for whenever someone posts news or articles about cyber security and all that it implies, it's other thing to have an AI system that can scrap all this data, and give you just the final results. As a human, you'd have to read everything, and most of the time there would be nothing new; the AI system can read everything in a matter of minutes, and retrieve just the new things.
- remote access that's more secure; many companies have workers that travel the world, or employees that work from home. They have to connect to the company's network, and they can do this with a VPN. But logging in with just a password can be a vulnerability; either the password is stolen, and someone can access the network from wherever they are, either the password is cracked, either the device the employee uses is infected, and some Man-in the Middle action can happen. Companies can use Artificial Intelligence to create a system where only the employee can log in, with a multiple factor authentication, for example.
Giving these, Artificial Intelligence in Cyber Security can help companies with the following:
- respond faster to breaches, as the AI system can detect them in no time;
- be accurate and efficient in cyber analyses;
- having the right tool to respond fast to cyber attacks.
A company can have a well-developed security system, that consists of multiple layers: network, application, endpoint and so on. Each part of the company is protected, but what if the hacker gets pass the firewall of a network? First of all, it wouldn't be detected by the human eye, and it can lead to access to various other antivirus systems or firewalls. The IT team can't have their eyes pinned to the monitors, trying to catch a new login, especially if the attacker used a method that can't be seen so easily.
With an AI system in place, the possible breach could be detected even before it succeeds. The AI system can be programmed in such a way that can point out any new login that has a new behavior, making it hard for the hacker to even get pass the firewalls. If he still manages to do it, of course the IT team would know immediately, if not before (there can be a notification of suspicious activity, which can inform the team right when the AI system detects someone trying to break the firewall).
In addition, the AI system can work 24/7, every single day, which can reduce the costs in the end.
As almost every single thing in this world, even Artificial Intelligence in Cyber Security has its disadvantages. Of course, they can be avoided, but you'll see below how.
The first disadvantage is the cost. Implementing a artificial intelligence system can be pricey. You need the proper men, you need the proper devices, and computer resources. The small companies don't usually implement this kind of system because of this, but also because they aren't targeted that often.
Even for the big companies the cost can be quite high, but they will amortize their cost -it will be an even bigger problem related to cost if there will be a data breach.
The second disadvantage is related to the training of the employees. In case you want to make use of your existing IT team, you should be sure that they will know the programming languages required for implementing an AI system. The system has to be built with no flaws, or at least as little flaws as possible. Otherwise, it can become a tool for the hacker, who can stir up things and make the AI system work for him.
From this part above, we can come up with another disadvantage: cyber criminals can use the AI to launch more sophisticated attacks. They can even go up to the point of impersonating the CEO or someone from the superior level. If there isn't a well-established system in place, things can go bad. For example, the employees should know that there shouldn't be cases where the CEO asks for money, especially large sums; there should be a process that is followed with attention, so that the employees won't fall for the scam. For example, immediately calling the superior to double-check what he/she just said.
All in all, even though there are some disadvantages, artificial intelligence in cyber security will still have more benefits. It's simpler to prevent and to invest in a powerful security system, than to pay the huge fines, to lose profit and to, worst case, lose the company.