Abnormal Stops the Most Sophisticated Lateral Phishing Email Attacks

One of Abnormal’s primary goals is to provide the maximum level of precision detection in order to prevent never-seen-before threats. This includes text-only payloadless attacks like social engineering attempts and account takeovers. Abnormal accomplishes this without requiring clients to manage and maintain a plethora of rules, unlike traditional secure email gateways (SEGs).

We are constantly improving our differentiated technology and superior efficacy when it comes to stopping email attacks. As a result, we have created and trained a new machine learning model that quickly eliminates identified false negatives, particularly for payloadless attacks like advanced socially-engineered emails to extract personally identifiable information (PII) of the victim.

The Abnormal detection engineering team has been busy at work, further enhancing our detection precision by decreasing our false-negative rate for lateral email attacks by a staggering 27.2%.

You might be wondering: what exactly is a lateral attack? Well, it’s defined as attacks sent from a compromised user to another internal employee with malicious intent. Lateral attacks are exceptionally challenging for traditional SEGs to detect as they usually have little to no visibility into internal email traffic.

A lateral attack consists of two components. The lure is the first and most crucial component, as the attacker must persuade the victim to take some action. This is accomplished by impersonating someone that the user will most likely trust.

The exploit is the second component. The victim is enticed to perform an action, such as opening an attachment with a malicious payload or visiting a URL that tricks the user into providing their valid network credentials.

In this example, Abnormal’s explainable from within the admin console explains why we detected and blocked this attack.

Once the attacker gains access to a corporate network, they will use various tools and techniques that elevate their privileges to achieve their objectives. For example, some will exfiltrate sensitive information to either sell to the highest bidder or put up for ransom. In other cases, the bad actor will deploy ransomware across the environment and demand a payment to access your information and bring your critical business systems back online.

Abnormal engineering determined that a heuristics rules-based approach is not overly effective at detecting lateral email attacks, given there are too many permutations. Instead, the engineering team took the following approach:

1. The engineering team did a deep dive into lateral attacks that identified clear attack trends within the data and retrained the detectors while directly addressing the attacker’s strategy. One notable example is students at educational institutions being targeted with fake job offers, enticing them to provide personally identifiable information or account credentials.

2. The analysis also surfaced sender-based behaviors like deviations in normal communication frequency and content that are strong signals of lateral phishing behavior. We have built a new feature-engineering platform that enables us to craft these new signals to adapt to bad actors' ever-changing tactics.

3. Retraining our detectors with the new signals to more effectively identify anomalous behavior in internal email communications has led to detection gains in lateral phishing and other aspects of the product, such as vendor fraud detection.

The net result is that the detection improvements effectively address the lateral attack strategies we have previously seen, and can accurately and automatically identify novel lateral phishing attacks that evade traditional email solutions. This allows us to provide our customers with the most effective email protection.

Abnormal is devoted to rapidly innovating to improve our detection capabilities, built-in automation, and processes to assist security teams in staying one step ahead of attackers. Updates to our revised algorithm prioritize key advancements, particularly those connected to detection enhancements, and provide a better overall experience for our consumers. At Abnormal, we stop the attacks that matter the most.

Want to learn more about how our advanced algorithms stop advanced attacks? Request a demo today.