Image-based Spam Adds Volume to the Inbox

Posted by Kevin McNally on 3 July 2009 | 0 Comments

Have you been wondering why you hear so many complaints about spam lately? You're probably thinking it's strange that spam is on the rise considering the Federal crack down on spammers and the increase in sophisticated spam-filtering techniques.

This phenomenon is not a global break-down in anti-spam solutions, but it is the result of a new take on an old technique called image-based spam. By definition, image-based spam contains its unwanted content inside of graphics. Embedding an advertisement's call to action in an image makes it extremely difficult for some spam filters to identify it.

The technique is apparently working: recent industry reports claim that as much as 30 percent of spam is image-based spam.

Because many anti-spam solutions rely heavily on text content to make a block/no block determination, these solutions are failing to identify image-based spam because of the way it presents itself. Image-based spam defeats traditional scanning techniques by visually portraying what would be normally a trigger word or phrase in a way that a text-based filter cannot interpret.

In addition to overflowing inboxes, which create frustration and stress, image-based spam is increasing the average size and bandwidth usage of spam. Images are larger than plain text messages – the average spam email was approximately 16K in the first quarter of 2006 and grew to 23K in the fourth quarter. The bottom line: spam is also wasting money!

In addition to the image, the unwanted messages have been relatively successful in bypassing many of the spam solutions used today when they contain:

• Text to fool stand-alone Bayesian engines or weak heuristic mechanisms
• Images sent as multi-frame animated GIF files
• An image that is a hyperlink to a Web site but often doesn't link to anything 
• Images embedded as a background picture rather than a body image 
• Images created using random lines or dots – a newer technique that can fool Optical Character Recognition (OCR) and fingerprinting methods
• Ransom note-styled messages in which each letter is an individual image to stop filters from looking for single images 

Be a spam-fighting hero

Remember the complaints? Abate frustration and be a spam-fighting hero with MX Logic Email Defense Service. Our solution integrates a patented Stacked Classification Framework that uses multiple filters to identify spam. By running email through a variety of identification techniques, our solution is not dependent on any single technology to detect spam. With IP Reputation and Real-Time Blackhole Filters, the Email Defense Service analyzes the origin of the message – not the content of the message – so image-based spam can be recognized and blocked like any other spam email.

The Email Defense Service uses both industry and proprietary heuristic filters to detect spam based on rule sets. While some of the image-based spam is resistant to heuristic detection, Email Defense has mechanisms that can catch it. Sometimes, image spam is resistant to statistical analysis through Bayesian filters, but after image spam is identified, the information becomes part of the Bayesian data set.

Finally, the Premium Anti-Spam Multi-Language Filter uses a global network of spam submitters and an intelligent fingerprinting system. This filter is used not only to identify a message as spam but also to detect variants of the original message. This filter has proved to be very successful at blocking the variations of images caused by randomized colors, dots or lines.

With Email Defense Service, image-based spam will no longer find its way into your inbox. With the Stacked Classification Framework – which incorporates more than two dozen spam-filtering techniques – on your side, you will receive less spam no matter what form it's in or how it's being sent.

For more information on MX Logic services, call Interactive Palette at 781-930-3199.