Using Machine Learning to Detect Command Line Anomalies

As we all know, cybersecurity is often a game of cat and mouse - attackers are always trying to outsmart us defenders. At Adobe, we face the same issues and concerns as all the other major companies. We must ask ourselves simple questions with non-simple answers: How do we ensure that all assets are protected? How do we ensure that our employees are secure from the outside threats? How can we mitigate future emerging threats? Attackers will always try to find the next unconventional attack that will bypass our security systems and our security mindset. In this case, how do we protect our self from the unknown? We believe machine learning techniques can assist us in this defense. This presentation will discuss one of our current machine learning innovations that is helping us detect anomalies in command lines. Command line interfaces are frequently used by users, system administrators and applications alike. Many applications launch console scripts to perform tasks, especially in cloud services where conformity in service environments is also helpful for security. When they can, attackers do like to leverage those native system capabilities. This presentation will discuss machine learning methods developed by Adobe computer scientists to help detect anomalies in command line scripts and calls to help prevent these types of attacks.