Machine Learning: A Protection Against Spear Phishing

Machine Learning Protects From Spear Phishing

How Machine Learning can help to protect your inbox from Spear Phishing

Email is the easiest medium for malicious attackers to target. Today, more than 90% of targeted Threats enter your Organization through Emails! 

The main difference between a common phishing email and a spear phishing email is that the common phishing email mimics a trusted organization and sends an email that requires the victim to log in into the account to collect data, while the latter is more targeted and personalized. Think of it as a custom-made attack that looks not only realistic but also is one on one, “person to person” email. Such an attack can be an email that appears to be sent from a person like a CEO to an employee (of target organization) making it very believable.  

It becomes essential for these more dangerous spear phishing emails to be detected and blocked by the cybersecurity measure in place. But what makes this difficult is that the more personalized and sophisticated a phishing email, the harder it is to differentiate the real from the malicious. 

Machine learning is one of the possible solutions to this problem. Andrew Goldberg explains that machine learning is an algorithm using powerful tool which requires usable data detect patterns and abnormalities. He also showcases the three main methods where the power of machine learning can be harnessed to detect spear phishing threats.  

Social Graph analysis –  

This method is based on identifying the typical patterns of communication that occurs within the company. A social graph is built through the process of observation of the information in the headers of every internal email. Frequency of the emails between the employees play a key role in mapping out the social graph. 

How does this help identify spear phishing attacks? 

Spear phishing attacks take advantage of the tendency of employees to obey orders from higher authorities such as the CEO and follow instructions given. The social graph analysis created by the machine-learning algorithms are able to detect abnormal emails based on the information in the connections in every email. These detected abnormalities are then presented to the recipient as a warning, thus reducing the risk of the user falling victim to the attack.  User communication profiling 

Like fingerprints, individuals have a particular style and tone of communication. These include something as generic as the use of emojis and abbreviations or something specific such as favourite or frequently used phrases. This becomes useful to detect malicious emails! 

What does this have to do with Spear Phishing?

A field called Natural Language Processing teaches computers to understand and model language. These techniques make it possible to extract identifying features from written text.  These can include phrases particular to certain areas and also criteria such as word choice, sentence structure and complexity of sentences. These markers can distinguish between different people when it comes to written text with the help of algorithms which operate by comparing the email to a model generated from legitimate emails.  

This method may not work for emails where a legitimate email is used as a template but the locations of a few links are changed. To combat this, it is best to use this method with a combination of other methods to get the best possible protection.

Email Structural analysis 

Apart from the regular details such as the sender, time, subject, body and attachment, emails also contain other information such as the IP address. Though IP addresses can be modified, the number of hops it takes to reach their destination can help in detecting malicious emails. Discrepancies in structure of the email such as missing headers or additional headers can also provide cause for suspicion.  

Machine learning creates a profile by observing the common structure of a users emails. This can help create a profile specific to each employee in a company. Abnormalities and possible malicious phishing attempts can be detected by comparing every new email to the profiles. Thus, suspicious emails can be flagged and the user can be notified. 

Protecting your emails is a way to protect your network and your business from spammers and malicious attackers. This is a global necessity for every organization as well as for individuals. As one of the leading IT companies that delivers Email Security from its private cloud Infrastructure, Logix has a core competency in securing over half a million mailboxes.  

We at Logix are dedicated to protecting your network from Advanced Malware (known and unknown Malware), Spear-phishing, Domain Impersonation/ Domain Spoofing, Zero day, Whaling, Targeted Email threats, Ransomware, Crypto ware, Business Email Crime (BEC) attacks with our multi layered, multi-tiered Security approach. We always use multiple threat intelligence detection & prevention tools.  

We specialize in configuring, migrating & supporting email security solutions for our valued clients. You can choose from the On-Premise or Cloud Based options. To know what will work best for your unique needs visit this link.

Leave a Reply

Your email address will not be published. Required fields are marked *

Continue to chat
Hello 👋
Let us know how we can help you!