Page 180 - Cyber Defense eMagazine April 2023

Fighting Bad AI with Good AI

So what can organizations—or the individuals who use email systems from both their offices and their
homes—do to protect themselves from these sophisticated AI-powered threats? The only valid strategy
is to utilize the same level of advanced AI technology in the cybersecurity solutions that protect their
email data.

The biggest challenge for companies looking to secure their networks is determining whether or not their
security solutions rely on sophisticated AI. The assumption is that all solutions leverage these methods,
yet it’s often not the case. Many traditional security solutions, including some of the most entrenched
brands in the marketplace, rely on the blacklisting of known IP addresses as their main line of defense
against malicious activity. Many established solutions were designed before this escalation of AI-based
cybercrime and are still catching up, as opposed to newer solutions that have been designed from the
get-go with AI-powered scanning tools.

It’s true that a great percentage of spam attacks come from IP addresses that have been previously
established as malicious. But blacklist/whitelist filtering doesn’t account for AI-based bots that read
through millions of internet references, interpreting contextual messages to create ultra-convincing,
natural language spoof emails.

Security solutions must employ the same level of AI technologies, leveraging these protocols to scan
inbound emails for red-flag keywords and phrases—the same way nefarious code will scan for data on
their victim’s hometown and business partners. These AI-powered solutions will either flag, block, or
quarantine questionable emails using keywords such as “credit card,” “invoice,” or “wire transfer.”

This is a true situation of fighting fire with fire. Only AI-based algorithms will be able to detect the cleverly
worded, accurately targeted, and naturally composed imposter emails that are soon predicted to reach
business environments. If an organization’s cyber security solution does not employ advanced AI tools,
they will be underprepared for the influx of well-crafted phishing attacks and malware.

Depend on Encryption

Another way to fight against powerful AI-related breaches is to implement the use of email encryption
throughout an organization, which keeps sensitive material from being accessible to hackers and AI bots.
Advances in encryption technologies have produced solutions that are simple to use, where encryption
can be automated to comply with regulations such as HIPAA or the GDPR, taking the burden of deciding
what content is sensitive and out of the hands of the individual user. Effective encryption allows protected
messages to be opened and decrypted with a single click, just as easily as a non-encrypted email.
Powerful solutions will auto-encrypt the recipient’s reply as well.

Effective encryption will not impact the user’s ability to breeze through their inboxes. Organizations should
look for an encryption solution that incorporates tokenization, a break-through capability that allows the
non-sensitive portions of an email to be read just as easily as any other email. The tokenized portions

180