Innovative Malware Detection Engine Powered by Artificial Intelligence
Malicious software or Malware is in the
business of doing harm to data, devices and people. Viruses, worms
and spyware are all variants of a very dangerous and quick growing
menace to your safety and that of your business. In recent years the
majority of devastating attacks were launched using notorious
Ransomware and Cryptojacking programs costing millions.
is excellent at its job – and you need to be excellent at yours to
defend against it. Traditional security systems and anti-virus are
successful at identifying existing signatures but are defenseless
against more recent highly malleable and malicious threats. Hackers
often just create new variants of existing malware or leverage
effective zero-day techniques against you and your business. How do
you defend against this constantly evolving threat? You need to
Artificial Intelligence technology has empowered Sangfor
to develop its own in-house malware detection engine to defend
against threats known, and unknown. Our dedicated R&D and Security
team - composed of data scientists, security analysts and white hat
researchers, are continuously developing & enhancing this engine in
conjunction with one of the most exciting technological advancements
of our time - AI.
“Engine Zero” has been created to ensure that
zero threats will affect your network and give you complete and
wholistic protection against zero-day vulnerabilities. Engine Zero
is only one of many malware inspection engines embedded in Sangfor’s
network security solutions, end point solution and Neural-X cloud
Traditional malware detection usually falls into the following categories: • Signature based detection:
Often used by
traditional anti-virus vendors, hash (MD5) of all known malicious files are
computed and stored in an anti-virus database. Each time a suspicious file
is inspected, its MD5 is computed and then compared against the other files
existing in the AV database. While this method is effective, fast and
industry-tested, it requires daily upkeep of a huge database of known
malware samples, often hundreds of megabytes, and daily endpoint updates.
While maintaining this constant vigilance is still valuable and effective at
combating known malware, it is still ineffective against fast-evolving
malware and malware variants. • YARA type Script Engine:
This script examines the suspected files/directories and
matches strings as they are defined in the YARA rules with the file. YARA
approach does a better job than AV in terms of covering more families of
malware however still falls short in detecting newly created malware. • Virtual Execution/Sandboxing:
execution and sandboxing is the process of detonating malware within a
controlled virtual environment and monitoring the post execution behavior.
The nature of malware presents several challenges to this method. Malware is
getting smart enough to recognize when it’s in a sandbox environment and
learns to avoid them in the future if the sandbox isn’t invisible (and they
usually aren’t). The other challenge with sandboxing is that it takes a long
time for inspection, and often not broadly deployed at all parts of the
network, or even at many organizations, allowing malware to bypass it .
While variants were notoriously difficult to detect in the past, we are
getting more accurate with our detection techniques. However, with increased
knowledge comes increased resource consumption and a pitifully slow
How Does It Work
Designed by dozens of Ph.Ds, scientists, security
analysts and white hat hackers, Engine Zero combs the entire suspected
file, categorizing it’s finds into multiple features. All agree that
machine learning is powering the way we protect ourselves and also how
technology will continue to grow, and Engine Zero employs several of
these effective and jaw-dropping techniques.
Based on years of
diligent security research on malware characteristics, Sangfor has
developed a supervised learning model. To train and ensure the accuracy
of this model, we then applied tens of millions of malware samples while
using advanced Artificial Intelligence capabilities to enable our engines to run
and teach themselves, expanding our capacity to discover unknown malware
and their families.
Engine Zero is not the only line of defense
within Sangfor’s security portfolio, including network gateways,
endpoint protections and cloud-based security as a service. Other
defense including threat intelligence, sandboxing, and botnet detection
capabilities are all working in concert to provide a comprehensive
coverage for malware detection.
Coverage of known and zero-day attacks. Our engine released in June of 2017 has proven itself able to detect high-profile malware such as BadRabbit ransomware, first seen in Oct 2017, without any previous signatures.
In recent tests our malware detection rate scored the highest in terms of accuracy, surpassing other vendors and open source alternatives.
This engine is very efficient and utilizes very little resource. Only such efficiency can provide malware inspection on the network gateway with very little performance impact.
Engine Zero Validation: Unmatched Protection for
While money-motivated attackers have been developing
and deploying more and more sophisticated ransomware, Engine Zero’s advanced
AI capability has proven itself highly effective against it. The higher our
success rate, the harder hackers need to work to create widespread
weaponized tools that are unique or variations on known malware. Click
here to learn more about the history & evolution of Ransomware. Engine Zero and Ransomware
Zero’s supervised training and Artificial Intelligence has proven itself the best
defense against ransomware offering:
Demonstrated to offer the best coverage for known and unknown ransomware
even without prior training (example: BadRabbit).
In 60,000 recent ransomware sample tests, Engine Zero scored the highest
among similar solutions.
: Customers using our
next generation firewall (NGAF) can use Engine Zero to detect ransomware at