Microsoft Research creates a JavaScript malware scanner
Microsoft Research creates a JavaScript malware scanner
Friday, 03 December 2010

Keeping JavaScript based malware at bay might just be a job for an AI-trained tool called Zozzle.



JavaScript is useful but the need to guard against malware hiding in deeply obsfucated code is a growing problem. Microsoft Research has invented a scanner that can detect malware using mostly static analysis - and has called it Zozzle.

Zozzle is a product of AI techniques. The researchers used a statistical classifier to scan millions of web pages for malware. The JavaScript has to be de-obfuscated first and then analysed for features that are characteristic of malware. The features are created using an abstract syntax tree. Zozzle hooks into the JavaScript engine to get the final expanded version of the JavaScript code the page contains.


At the moment the tool is in the research phase and there is no date set for release into the wild. Current performance is claimed as less than 1% false positive with a typical 2-5 millisecond processing time per Kbyte of code.

The researchers say that they can envisage it being used both within a browser context to protect against malware on the fly or in an offline context to classify and blacklist infected sites.

More information



The Weekly Top 10: eCommerce Development Resources

eCommerce, in particular Shopify and Bigcommerce, is the focus of this week's trawl through the web to bring you items of interest that we've not had chance to cover in our daily attempts to cove [ ... ]

DeepMind's Differentiable Neural Network Thinks Deeply

Neural networks are great at reacting to complex data, but not so good at the sort of slower thinking it takes to solve a problem like getting from A to B using a subway or unscrambling a sliding bloc [ ... ]

More News

Last Updated ( Friday, 03 December 2010 )

RSS feed of news items only
I Programmer News
Copyright © 2016 All Rights Reserved.
Joomla! is Free Software released under the GNU/GPL License.