AI in Cyber Defense: The Good, The Bad, And the Ugly
Updated: Jun 1, 2021
Cyber security is an ever-shifting landscape, and cyber criminals are continuing to become more sophisticated to shift with it. Because of this, many cyber professionals are trying to figure out how to mitigate risks as much as possible. Accelerating the development of digital tools in the process.
Some of the key tools in the works are that of Artificial Intelligence (AI) and Machine Learning (ML), where experts are dubbing these processes as the answer to helping stall exploits from the likes of data leakages or hacks. But how can we be sure that AI is truly trustworthy? And should we be meeting these technologies with open arms or a fine-toothed comb? Well, we decided to take a look at the full picture, weighing up the pros and cons of AI/ML and where it stands in being a fundamentally sound solution.
First, let's take a look at the positives:
1. AI can handle a LOT of data
Most companies manage a large amount of data daily. This is especially true when it comes to businesses spread across a range of locations, multinationals or even sponsored operations. Therefore there needs to be some serious capacity available to maintain normality. Even in the personal device space, one can receive hundreds of emails a day that may contain potentially harmful content. A robust AI/ML algorithm can detect threats before they happen, and file through a sea of data efficiently. Therefore, simplifying and automating the process of scanning an email and searching for inconsistencies, for example.
2. Unknown threats are easy to identify
During 2020, Statista projected the development of new malware detections would rise to over 700 million by year's end, that is an estimated 1.9 million new types of malware developed and detected daily. However, when cyber security is automated, the focus is on constantly registering changes in the network to identify new threats.
3. Available 24/7
The typical work day of a human lasts around 6-8 hours daily, and very rarely does a human work 7 days a week. The work life balance is crucial to ensure people are in the right headspace and get enough time with their friends and family, we also tend to get bored pretty easily or get overtired when we are burning the candle at both ends. Unlike humans, AI machines can work 24/7 without complaint, they also aren’t too phased if they don’t get leave around holiday seasons.
4. Helping in repetitive jobs
The human work life is often repetitive. On a daily basis, we will likely be sending thank you emails, spell checking forms or verifying documents, and these are just a few of the tasks we see day-in and day-out. These tasks are often time-consuming or unmotivating, and this is another area where AI can help us out. A lot of these processes can be automated, freeing up workers for more creative and productive operations that demand attention.
5. Reducing Human Error
Like it or not, we as humans need to make sure we are looking after ourselves to do our jobs effectively while also monitoring the ever-growing problem of mental health and a bad work-life balance. This leads to a range of different emotions such as anger, mental fatigue or indifference. The trouble is that these things are rather common and as a result, human error is extremely prevalent. We make mistakes, and we make a lot of them. AI, however, does not make the same mistakes when programmed properly. When it comes to AI, the decisions are taken from previously gathered data and run through a series of algorithms to not make the same mistake again (hence the machine learning thing). Although the same can be said for humans in some regard, we aren't quite as quick or consistent.
There are many benefits to implementing AI into your workplace. It can do a decent job of making your life easier when it comes to detecting and reducing errors, availability, and identifying threats however, it is important to look at both sides of the coin to ensure we are getting too far ahead of ourselves.
1. High Cost of Creation
AI machines are extremely costly to make and maintain. In order to meet the pace of latest requirements, AI needs to be updated daily and for it to remain feasible there can be an exceptionally high cost when updating hardware and software. It also takes time. There are only a few ways currently that AI emulates people, but one of these is that it has to make a few hiccups first before becoming completely reliable. These machines also need repair and maintenance which comes at a high price ticket. They take time to develop as they are very complex machines, especially when on their maiden voyage.
2. Lack of out of the box thinking
There is certainly one thing that we do better than machines, and that is out of the box thinking. When it comes to AI, they can only do what they were designed to do. They work in algorithms and equations in order to work, so chances are if you are wanting your accounting AI to write a unique screenplay for a blockbuster film, cast the actors and juggle ice cream, or write creative copy for the latest advertising campaign, chances are it will likely fall quite far from expectations. When the AI is tasked to do something that is out of their core programming or purpose, they tend to crash or give random outputs which could do more harm than good.
3. Cyber threats evolve
Sadly, protective and preventive AI goes through intricate research before becoming available in a fully functioning format. Most researchers will release such projects in an open-source format and have extensive research or development notes so that everyone can access them, and this includes hackers. Hackers can then manipulate the code (likely still improving the product slightly to avoid detection) and add snippets to it to prepare for a future cyber attack, confusing the code.
4. Cybercriminals may know AI well also
You don't have to go far to learn about how AI works and what code is. This information is freely available to everyone. In fact, with enough dedication one doesn't even have to be overly IT-savvy either. Cybercriminals can easily get a hold of the cybersecurity software and develop AI-proof hacks that may bypass the security procedures or expose vulnerabilities in the code. It is important to remember cyber criminals can use ML to understand ai-based cyber solutions or pollute malware to make it seem harmless.
When looking towards potentially implementing AI-based solutions for you or your business, one must assess every possible aspect. It is clear to see that AI is quickly becoming the way of the future. There are definite benefits to AI-based solutions, but there are some rather big hurdles that may stop AI from being the be-all, end-all solution when it comes to cyber security.