By Roark Pollock

AI in Cybersecurity: Why Do Cyber-Hacks Still Succeed?

This is my 3rd blog on artificial intelligence (AI) in cybersecurity, coming in the wake of our recent announcement that Ziften Zenith has added Advanced anti-virus (AV) built on a foundation of AI.

Previously, we took a look at how machine learning-based cybersecurity systems operate differently from traditional, signature-based antivirus software. We also discussed the preponderance of so-called “fileless” attacks, and why AI-based cybersecurity is the best defense against them. We’re excited to bring these cutting-edge new defenses to our existing endpoint protection platform and our customers.

In this final entry, let’s address a seeming paradox in this new era of AI-powered cybersecurity:

With such sophisticated methods for cyber-defense, why are attacks
still happening?

Now more than ever, it seems, organizations ranging from small tech companies all the way up to entire countries are vulnerable to hacks, breaches, and other attacks on their digital infrastructures. If AI is being used to defend their networks and endpoints, what’s going wrong?

Separating the Wheat from the Chaff

The first answer is simply that AI-powered cybersecurity is not as ubiquitous as it should be throughout today’s enterprise organizations. Companies and especially people are slow to change. Thus, enterprises have failed to move from traditional, signature-based AV systems to newer endpoint protection platforms built on AI. And this continued sole reliance on signature-based AV is exposing enterprises to new fileless attack vectors. As we learned in the previous post, “The 2017 State of Endpoint Security Risk” report by the Ponemon Institute, found that fileless attacks were included in an estimated 77% of successful data breaches worldwide. Without AI-based endpoint protection, enterprises will continue to fall prey to these fileless attacks.

Additionally, not all AI cybersecurity tools are created equal. Antivirus programs that advertise the use of AI often fall short in bringing its full potential to their defenses, failing to defend against many zero-day threats that should be prevented with ease. This is partially due to the fact that machine learning talent is scarce, and companies are finding it difficult to hire data scientists and engineers capable of developing sophisticated machine learning products. But also, for large, well-established cybersecurity firms that can find such experts, the inertia generated by years of designing traditional AV software is hard to overcome.

AI-Based Cyber-Threats

As we mentioned in a previous post, malicious actors have been quick to add the capabilities of AI to their own arsenal. Simply put, the same techniques that are making cyber-defenses stronger are also empowering cyberattacks. Machine learning algorithms have been trained to write code to evade the scans of traditional AV, while other AI tools trained to recognize valuable data are sweeping the internet in search of personal information at lightning speeds. In some cases, hackers are using AI-powered chatbots to create mass phishing attempts, or even to sow discord across social networks and organizations.

Frighteningly, these techniques may be just the beginning. Many cybersecurity experts are warning of the potential of “deep fakes” – fake video clips created by artificial neural networks that seem incredibly authentic – to spread harmful misinformation or trick users into compromising their digital defenses. Watch this BBC Click episode on YouTube to see these “deep fakes” for yourself. In the future, deep fakes disguised as official company materials may be used to trick a firm’s employees into divulging passwords or other confidential information.

Human Error, Human Ingenuity

Not all successful hacks are high-tech in nature. Sometimes, all it takes for a breach to occur is something as simple as leaving a computer unlocked or sending an email to the wrong address. Powerful as it may be, AI isn’t capable of preventing many forms of human error or carelessness or preventing a thief from swiping your laptop or smartphone. Even with AI-based defenses, it’s important to stay vigilant, and to break some of the bad habits that can open the door to hackers. Here are a few things to keep in mind:

  • Never assume you’re safe on public Wi-Fi
    Most public networks lack the security features to prevent other users on the network from spying on your activity. At the very least, avoid signing into sensitive accounts while on a public network.
  • Don’t use the same passwords all the time
    Tempting as it may be, resist the urge to use the same passwords (or close variations of them) for all of your accounts. Likewise, if you must store your passwords digitally, don’t put them all in the same place. Better yet, invest in an encrypted password database and select a very difficult to crack password for this one database.
  • Be wary of unfamiliar emails
    While this one may seem obvious, you’d be surprised at how often people click on links or attachments without looking at the sender. Keep in mind that we can be especially careless when tired, or deep in focus on another task.

Learn more about how Ziften is expanding its Zenith endpoint protection platform with artificial intelligence:

Get the Featured General General Here