Friday, October 20, 2006

Why OCRing spam images is useless

Nick FitzGerald forwards me another animated GIF spam that takes the animation plus transparency trick I outlined in the blog post A spam image that slowly builds to reveal its message to a new level. And it shows why spammers will work around OCR as fast as they can.

Here's what you see in the spam image:



Looks simple enough until you take a look at the GIF file that actually generated what you see. It's animated and it has three frames:





The first image is the GIF's background and is displayed for 10ms then the second image is layered on top with a transparent background so that the two images merge together and the image the spammer wants you to see appears. That image remains on screen for 100,000 ms (or 1 minute 40 seconds). After that the image is completely blanked out by the third frame.

My favourite touch is that it's not the entire image that's transparent, not even the white background, but just those pixels necessary to make the black pixels underneath show through. If you look carefully above you can see that some of pixels appear yellow (which is the background color of this site) indicating where the transparency is.

That is darn clever.

8 comments:

Original Syn said...

It's a snake eat tale situation, sooner or later anti-spam companies will come up with a way to OCR these images, this could simply be defeated by merging the frames and then runnings the software. And when these tactics no longer work spammers will find a more novel way of absuing some technology.

Anonymous said...

Simple solution: block emails with gif attachments.

I can't remember the last time I received a valid gif attachment.

Ron said...

There may be a simple solution that doesn't involve blocking GIF entirely. If the GIF file itself can be inspected for multiple images then you should be able to send to quarantine or OCR based on that check. I would think that there must be something in the header and/or some delimiter between the images.

Bronskrat said...

don't block gif's, block GIF 89a's (a for animated). It's in the header.

Nick FitzGerald said...

original syn "...sooner or later anti-spam companies will come up with a way to OCR these images..."

Actually, there has been an OCR plugin for SpamAssassin since early this year.

"...this could simply be defeated by merging the frames and then runnings the software."

It's a bit more complex than that. Read John's other blog and TSC http://www.jgc.org/tsc/ entries about image[-only] and animated GIF spam, paying attention to the initial emergence of animated GIFs with "noise frames" and such. You may also get a feel for what is involved by reading the "history" comments at the top of http://antispam.imp.ch/patches/patch-ocrtext

Nick FitzGerald said...

mkaatman said "Simple solution: block emails with gif attachments."

Not only simple, but simplistic. This may work for you, but in general will have a fantastic false-positive rate. You may not care, but for very good reason anti-spam developers are desperately sensitive to their FP rates.

Nick FitzGerald said...

ron said "If the GIF file itself can be inspected for multiple images..."

It can be...

...then you should be able to send to quarantine or OCR based on that check."

...but Incredimail users are prone to including animated GIFs, as are various webmail (and other) users who like to include those (small) animated sprite/avatar/etc images in their signatures. Again, it's not many folk in total, but enough to prevent serious anti-spam developers from adopting such a simple rule because of what it will do to their FP rates.

Paul McNamara said...

Interesting wrinkle. John had a few things to add in an interview he had with me, including the fact that he and Nick FitzGerald saw this one coming:

http://www.networkworld.com/community/?q=node/8977

And Nick tells me that if it's fresh spam you want, it pays to be located in New Zealand:

http://www.networkworld.com/community/?q=node/8993