Aren’t you sick of so-called anti-spam specialists banging on about how good their solutions are. I mean, if their products are so good, how is it that spam has risen from something like 80 percent of all email traffic to over 90 percent in the past few months?
It’s easy to criticise, and the truth is things would be a whole lot worse if we just left the spammers to it. Indeed, when it comes to tackling the growing problem of image-based spam, there’s some pretty clever stuff out there.
Most traditional filters are unable to detect and block image-based spam effectively because those clever spammers manage to make thousands of variations, each time altering the image by a tiny fraction, so that it can sail clean through. So what to do?
German web hosting firm Strato thinks it has the answer with its fingerprinting technology.
Developed in partnership with the Institute of Computer Technology at the Humboldt University in Berlin – boffins, white coats, you get the picture – and a certain Professor Scheffer of the Max Planck Institute – über boffin – this technology works by assessing the percentage of a certain colour in a specific tone, or by “the composition or structure of individual graphics”, according to Strato. It then decides the probability of that spam mail coming from the same sender as another, and is therefore able to block it.
Another key feature of Strato’s anti-spam offering is called social graphs, which looks at the relationships between senders and receivers and decides if a message being sent to your inbox is likely to have come from a certain IP address. If several messages are sent back and forth between two addresses, for example, this shows they are not likely to be spam.
Another firm at the forefront of the fight is Tumbleweed. Although traditionally associated with outbound content security, the firm has made impressive strides with the inbound piece, boasting the industry’s lowest rate of false positives, according to chief executive James Scullion. Its spam filters use a similar technique to Strato’s, which it calls Adaptive Image Filtering – well, you’ve got to differentiate, haven’t you – and includes a “fuzzy matching” technique that allows for minor changes in the images.
It also features other technologies such as Recurrent Pattern Detection, which concentrates on identifiable patterns in spam outbreaks. And yes, you guessed it, there’s also some form of AI here, in the firm’s Intent Based Filtering, which analyses the language in the messages to detect if they might be spam.
IT security vendors will always be reactive and nowhere more so than in the field of anti-spam, but the firms that are thinking slightly differently about the problem may offer the best chance of saving your email system from meltdown.
