ZOZZLE: fast and precise in-browser JavaScript malware detection

JavaScript malware-based attacks account for a large fraction of successful mass-scale exploitation happening today. Attackers like JavaScript-based attacks because they can be mounted against an unsuspecting user visiting a seemingly innocent web page. While several techniques for addressing these types of exploits have been proposed, in-browser adoption has been slow, in part because of the performance overhead these methods incur. In this paper, we propose ZOZZLE, a low-overhead solution for detecting and preventing JavaScript malware that is fast enough to be deployed in the browser. Our approach uses Bayesian classification of hierarchical features of the JavaScript abstract syntax tree to identify syntax elements that are highly predictive of malware. Our experimental evaluation shows that ZOZZLE is able to detect JavaScript malware through mostly static code analysis effectively. ZOZZLE has an extremely low false positive rate of 0.0003%, which is less than one in a quarter million. Despite this high accuracy, the ZOZZLE classifier is fast, with a throughput of over one megabyte of JavaScript code per second.