Cybersecurity in the Age of AI

Data is the new gold. We are becoming more and more involved with it in our daily lives. Whether it be the Internet of Things, self-driving cars, smart AI, or even the up-and-coming Neuralink, all of them rely on the flow of data to operate.

Table of Contents

With increasing reliance on data to drive day-to-day operations, this begs the question: how do we know that all of this data is secure and safe from the hands of hackers?

This is a question that goes mostly unanswered and will be the main focus of my talk today.

The Rising Threat

With cyber attacks and data breaches escalating exponentially in size and severity, neglecting cybersecurity is an issue that empowers hackers beyond anyone’s wildest imagination and a mistake that will cost us colossally in the long run.

Now, in order to truly go beyond the human, in order to truly move forward as a species, we must ensure that the tools we use to do so are safe and secure for the AI we want to use.

AI: A Hacker’s Best Friend

I’m sure you’ve all seen the advanced AI models and their magical capabilities—OpenAI, ChatGPT, Google’s Bard, GitHub’s Copilot, you name it. Some of you might have even used it to increase your productivity tenfold.

Ladies and gentlemen, what if I told you that you were not the only people using these to increase your productivity?

As it stands, AI is a hacker’s best friend, and it proves a surprisingly capable assistant in manipulating both the human and the computer.

AI and Malware Creation

This brings me to my first point, which is AI’s ability to write malware and exploit code.

Now, while AI models are centuries, even millennia away from actually replacing programmers, that’s not to say that they aren’t efficient at writing code—namely, dangerous code that can be used against internet infrastructure and systems worldwide.

To prove my point, I asked a nondescript AI to write me malware.
Disclaimer: do not under any circumstances do this. This is for educational purposes only.

That being said, this is its response, and here is what it returned:

A polymorphic self-encrypting virus.

For those unfamiliar with computer terminology, this is a virus that can change its appearance at will, which makes it very difficult to track and basically go undetected by most antivirus solutions today.

Now, keep in mind, ladies and gentlemen, that this was created in seconds by a bored high schooler. Imagine what a hacker with more ingenuity and more time on their hands can do.

Neglected Cybersecurity

Is this terrifying you yet? It gets worse.

Consider the fact that most companies nowadays neglect cybersecurity—yes, even the major ones. For example, Toyota in May 2023 released the fact that they had been the victim of a decade-long data breach which compromised millions of user accounts.

With cyberattacks increasing rampantly and becoming much more savage, and companies stagnating and refusing to improve their cyber policies, this stands to lead to a cataclysm in which your data is the victim.

AI and Social Engineering

The intrinsic value of AI to a hacker is not in its ability to write bad code. It is actually in its ability to manipulate people.

Social engineering is a class of attack that is becoming increasingly more prevalent, with 41% of major breaches happening due to this type of attack. It involves a hacker manipulating a human and coercing them into doing their bidding.

AI can help this in a very efficient manner.

Example: John Doe

Let’s take a look at this from the hands of a hacker.

Here is our victim: John Doe. We want his bank details.

Step 1: Compile a list of public information.

He’s 23
He’s a junior analyst
He’s a hustler
He’s struggling with his finances

We feed this to an AI model, and it returns a very convincing script for an email designed to trick him into clicking a malicious link.

This email prompt uses time urgency and consequences to manipulate John Doe.

The Harsh Reality of Phishing

Now, you might think: Oh, this is just a simple scam email. I won’t fall for this.

Two points prove otherwise:

It is becoming increasingly harder for people to tell the difference between AI-generated content and real content.
A 2021 study by IT firm AAG found that the success rate of phishing jumped from 18% to 51% when the attacks were personalized.

That’s a coin toss probability. Do you want the integrity of your data reliant on a coin toss?

Why Your Data Matters

Some of you might roll your eyes at the mention of your data, thinking it’s completely separate from who you are. But let’s contextualize this.

Raise your hands if you:

Have bought something online
Have private documents on your computer
Have done something private online
Have a smart device like Alexa or Google Home

The majority of you said yes. All this data acts as strings a hacker can pull to control you like a puppet.

As a result, the security of your data matters.

Fighting Back with AI

Now we’ve seen how hackers use AI to wreak havoc. But the important question is: how do we fight back?

The answer is simple: fight fire with fire.

Use AI to analyze long terms & conditions and make informed decisions about your data.
Use AI to detect social engineering attacks, since these models are good at identifying what they generate.
Programmers can use AI for boilerplate code, freeing up time to focus on security, scalability, and efficiency.

Combined, these make for a robust cyber strategy that can secure your data effectively.