MAGIC predicts malware evolution

“Working with Cythereal, we have developed systems that identify malware families, determine the lineage of each family, and predict the future evolution of a family. These systems require a deep understanding of the common characteristics shared by different members of the same family so that we can identify them as all belonging to a common family. Traditional features such as n-grams of code don’t provide this understanding, because they can easily be changed by a malware author. Cythereal’s Malware Genome concept solves this problem by identifying core characteristics of malware that don’t change when the code changes. The Malware Genome is robust to many kinds of code transformations, such as those performed by polymorphic malware. Our tests have shown that our methods are significantly more accurate using [Cythereal’s] malware genome than using raw code.” 

~Dr. Avi Pfeffer, Principal Scientist | Charles River Analytics