How hackers and cybersecurity experts are both using AI to outdo each other
Even as we remain focused on the data privacy threat posed by the world's biggest tech giants, what has largely gone unnoticed are their efforts at protecting our data – and even money – from hackers looking at disrupting our digital lives.
The likes of Google and Microsoft, who have, time and again, found themselves being called out for their widely reported invasive policies, and alleged attempts at collecting information about their users, are also responsible for keeping sensitive data – such as our bank details and other personal files – out of the clutches of hackers.
But with the cybersecurity landscape evolving at a rapid pace, and hackers using newer, more innovative ways to bypass legacy security systems, these tech giants are now looking at Artificial Intelligence (AI) to help them stay a step ahead in the race.
As Bloomberg explains in a recent report, Google, Microsoft and other big players in the industry have started to move away from simply using older "rules-based" technologies designed to respond to specific types of cyber attacks, and are instead deploying dynamic machine-learning systems to figure out patterns and stop suspicious activity on their servers.
Speaking about the issue, Dawn Song, a professor at the University of California at Berkeley's Artificial Intelligence Research Lab said, "Machine learning is a very powerful technique for security - it's dynamic, while rules-based systems are very rigid... It's a very manually intensive process to change them, whereas machine learning is automated, dynamic and you can retrain it easily."
The newer technology was seen at work recently when Microsoft's Azure security team used AI to track suspicious activity in the cloud computing usage of a large retailer.
Because the Azure system employed machine learning as part of their security system, Microsoft was able to figure out a suspicious pattern during the login process, helping it foil the attack before it got too late.
But, as is with most things, the game isn't as simple as it appears.
Hackers too have started to use AI tools to bypass advanced technologies being employed by these tech giants.
And at the core of such attacks, is troves of random data being thrown at these security systems to trick them.
From the time when multi-billion dollar machine learning systems such as Azure are built and trained, to when they become completely functional, attackers are now using malicious data to add slight biases into the system that could later be exploited by an attacker to gain entry into our bank accounts, or carry out fraudulent credit card transactions.
From large-scale security systems to self-driving cars, as machine learning finds deeper inroads into our modern way of life, it is thus becoming more and more crucial that cybersecurity experts start plugging these loopholes in systems that the industry is looking at to take us into a new era of security in the digital age.
Else, these machine learning-based tools could soon end up being the biggest mistake in the history of cybersecurity.
Also read: Why India urgently needs a cybersecurity law