Why AI May Not Be the Ultimate Answer to Cloud Security
February 23, 2021
26         0

by Stephen Kanyi

In 2019, 293.6 billion emails were sent every day. Is your company security strategy prepared to handle this upsurge in data? More importantly, with the continuous adoption of cloud technology, is your IT team capable of protecting your data?

As the internet expanded and with it the cloud, businesses realised that their IT teams weren’t capable of protecting them from attacks. The internet significantly increased the attack surface and at the same time hackers were getting smarter and used more sophisticated tools. Businesses and institutions at large needed to arm themselves with new tools to survive in this increasingly hostile cloud environment.

Enter AI.

Beginning as mere fantasy in the back pages of science fiction novels, AI is central to the modern economy. A combination of AI and Machine Learning is today helping shape a future only a few of us could have imagined a few years ago. From Google’s AI-Powered Predictions, AI autopilots to self-driving cars, the present and the future is AI.

With the increase in volume and complexity of data in the cloud it was inevitable that AI would be employed in the next evolution of cyber security. Company executives are turning to AI/ML powered cloud security in droves. A report by Capegemini found that 69% of executives felt that AI is essential for responding to cyber threats, while Cisco reported that 50% of leaders identified AI as a priority. 

The hype of AI in cloud security has recently grown to fever pitch. It’s the buzzword in every business meeting and conference. Business executives and consultants are all pushing their employees to embrace AI as the silver bullet to solve virtually every problem in the organisation. From sales, HR and more relevantly cloud security. And no one can blame them, we are all witnessing the raw power of AI in virtually every sector of our lives.

However, as people who deal with the nuts and bolts at the back-end we have to be realistic about our optimism. We have to take a closer look on the actual reality in cloud security. Is it really the silver bullet? Indeed, a closer inspection into the real capabilities of AI in cloud security shows that it is not as effective and powerful as people think it is. Let’s take a slightly deep dive into the promises and real capabilities of AI in cloud security as it stands today.


Here we try to separate the hype from the facts. While AI is truly powerful it does have its limitations. In fact, Bernard Brode, a product researcher at Microscopic Machines in a brilliant post about AI in cloud security says “…as far as I’m aware, there is no such thing as an AI-driven cloud security system.”

While this statement may seem to be an overly negative view of the situation it does say something about the real capabilities of AI cloud security. Let’s look at some of them.

Big Data Processing

This is perhaps the most developed and hence the most promising use of AI in cyber security. With the volumes of data having mushroomed in recent years, AI is the only tool that can effectively sift through all this data to identify attack patterns.

These kinds of systems are especially effective in identifying unusual activity and block the user’s access before s/he can do any real harm. This is very important in the overall management of employees’ access to files and systems.

Event Prediction

This is the next step in trying to create a truly ‘intelligent’ system. It involves applying AI’s powerful pattern recognition capability and extending it to a system that is able to make meaningful predictions in cloud security.

These sorts of systems are still somewhat underdeveloped. However, they do show a lot of promise if engineers manage to get them running as they should. Predictive analysis can be divided into two:

  1. AI systems are given information about the kind of systems that exist in the wild and the types of organizations being targeted. It will then try to make a prediction based on this data. This is more akin to the way neural links are used to produce car insurance quotes based on comprehensive data.
  2. Another involves “white hat analysis”, a term used to describe ‘friendly hacking’. Basically, companies hire professional hackers who attempt to break into their systems with any means possible. An AI system will then recognise patterns from these attempts and be able to protect the system in the future from similar attacks. It is however important to note this method is “still some years of “, as Brode puts it. He also adds that it creates possible danger as it may be arming hackers with a powerful set of tools.

Automated Response

This is where the AI systems are empowered to directly act on the systems they operate on. While many companies do tout their AI systems as being able to “intelligently respond to attacks in progress” as Brode puts it, the reality is that such claims are overstated.

First, the real capabilities of such systems are not really “intelligent”. They are simple coded rules that administrators use to point them to the right direction and make automated suggestions.

Even when deployed on systems, these AI algorithms are only capable of blocking certain users if say they attempt to access restricted files. In such a situation a rogue AI may cause a lot of harm by locking out hundreds of users.


That said, it does not mean that AI is totally useless in cloud security. As pointed out earlier, if used correctly AI can help security professionals identify attack patterns and also alert them of potential dangers. That is however, as it stands, the limit to its power. Anything over that is mostly hyperbole and it is important that executives know this distinction for the safety of their companies’ data.

subscribe for YouMedia Newsletter

Leave a Reply

Your email address will not be published. Required fields are marked *

subscribe for YouMedia Newsletter