ChatGPT: Taking Phishing Attacks to a New Level

Contact Us

ChatGPT: Taking Phishing Attacks to a New Level

Some industry analysts are heralding ChatGPT as the solution to many cybersecurity problems. After all, hackers are using artificial intelligence to boost the scale and frequency of their attacks to unprecedented levels. Organizations are recognizing the need to leverage AI and machine learning to fight back against the onslaught.

It’s important to not lose sight of the first half of that equation. Hackers are already using AI, and ChatGPT gives them a new, powerful tool. Beyond AI-supported password guessing, advanced persistent threats and automated penetration testing, ChatGPT opens up sophisticated capabilities for the least sophisticated hackers.

For example, ChatGPT enables phishing at a scale that wasn’t possible before and removes some of the key indicators of phishing. Organizations will have to retool their processes for detecting and neutralizing phishing attacks.

Interested In Learning More?

Fishing hook on top of a keyboard.

The ‘Human Firewall’ No Longer Works

Historically, phishing has been a manual process executed primarily by foreign entities. Threat actors would write phishing emails in their native language and run the text through an online translation tool. The results often had grammar, syntax and logic errors making low-level phishing emails easy to spot.

Another telltale sign of phishing has been the lack of an immediate response. Phishing emails are usually sent on timers to coordinate with the recipient’s time zone. They don’t have a human sitting behind them waiting for the victim to respond.

ChatGPT eliminates those indicators. The software can generate personalized emails with well-written text that accurately reflects the knowledge and style of the alleged sender. It can also generate an immediate response that looks as if it’s coming from the supposed sender of the phishing email. Users will have a hard time detecting fraudulent emails, and security awareness training will become less and less effective in creating a “human firewall” against phishing attacks.

Person typing on a computer with blue lock graphics all over the screen.

More Security Controls Needed

Organizations will have to lean more on their email systems to identify possible phishing emails and alert users. One effective technique is to give the user an indication of where the email originated. If an email goes from one person to another inside the organization, the email system adds a green banner saying that the message is likely safe because it’s internal. If an email comes from outside the organization, the email system adds a red banner saying that the message requires additional scrutiny.

Let’s say a user receives an email from her boss asking her to transfer a large sum of money to an external bank account. If the email has a green banner, the user can feel fairly certain that it came from her boss — although she should still phone her boss to verify. If the email has a red banner, she can assume it’s fraudulent.

The key here is to have a company such as DeSeMa set up the rules and tune the email system to accurately identify external messages. There’s also another element that comes into play.

Phone screen with four email notifications, all with red warning symbols next to them.

Spotting Behavioral Anomalies Becomes Critical

For the flagging method to work properly, you must know if an account has been breached. Otherwise, fraudulent messages could be generated from inside the network.

DeSeMa helps organizations utilize tools that detect and analyze behavioral anomalies to spot compromised accounts. If an account starts operating oddly or is being accessed from odd locations, the security controls should flag any emails from that account as potentially fraudulent. It should include a banner that says, even though this email appears to come from the right person, it didn’t follow that person’s usual behavior pattern.

In the age of ChatGPT, traditional methods used to identify and block phishing emails are no longer effective. Our next post will discuss how ChatGPT enables more adaptive malware to gain a foothold in your IT environment.

Get Started Today!

Contact Us