Researchers and IT managers are confirming security vendors' claims that spam levels have spiked in the past month - some say by as much as 80 percent - and show no signs of decreasing.

"There are enormous amounts of spam; it's shot up like crazy since the beginning of October," says John Levine, president of consulting firm Taughannock Networks and co-chair of the Internet Research Task Force's Anti-Spam Research Group. Levine operates a number of e-mail addresses that aren't filtered for spam. "Earlier this year I was seeing about 50,000 spam messages a day, now I'm seeing 100,000," he says.

Levine's assumption is this spike in spam levels is a result of a new generation of viruses and zombies that can infect PCs more quickly and are harder to get rid of. In its October intelligence report, messaging security vendor MessageLabs says the spike is largely caused by the dominance of botnets, networks of compromised PCs that are turned unwittingly into spam servers. The report also identifies two Trojan programs, Warezov and SpamThru , that probably are contributing to the rise in spam.

Others say a new breed of spam messages called image spam - messages with text embedded in an image file that evade spam filters, which can't recognise the words inside the image - is responsible.

At North Shore-LIJ Heath System, a network of hospitals based in Great Neck, N.Y., with about 12,000 e-mail users, there has been an 80 percent increase in spam received in the last 45 days, says system architect Steve Young. Most of this is image spam, the majority of which is pump-and-dump scams, in which spammers purchase a penny stock, promote it through e-mail, then sell it at a profit. Most appear to come from Europe, Young says.

"We got slammed with a 50 percent increase [in spam] in one day. For the past year and a half none of my users ever got a spam message; in that first 48 hours [of image-spam blasts] there were 500 calls and over 1,000 complaints from users," he says.

Young called BorderWare, his e-mail security vendor, to ask for help. The company enrolled him in a beta program for its new technology designed to block image spam, which Young says is working. "We blocked 7,000 image spam messages in the first day" of trialing the technology, he says.

What's made image spam so vexing is that spammers have learned to represent words in an image that are distinguishable to the human eye because of the way people recognize images, but can't be understood by computers in the same way, says Andrew Graydon, CTO of BorderWare.

Spammers are "banking on eye recognition, and so many of the solutions out there only deal with text analysis," Graydon says. The company's new technology, called Intercept Image Analysis and set to be unveiled this week, analyses image spam and comes up with a characterisation of the message that tracks 30 pieces of information about it that mimic the way people visualize.

Of course, as vendors come up with new techniques, spammers do, too. Image spam began popping up a few months ago, and some security vendors responded with products that create a "fingerprint" of the message and match that against new incoming messages. Then spammers began randomising image spam so that each message was slightly different from the last, therefore evading fingerprinting technology.

"On a scale of 1-to-10, I would rate image spam as an 8" in terms of how troublesome it is, says Paul Judge, CTO of Secure Computing. "This is because spammers have leapfrogged from hiding text within other text to now moving it to a place that is unreachable by most anti-spam systems."

Secure Computing is touting its TrustedSource Message Reputation fingerprints, which take a snapshot of a message identified as spam and assign a correlating reputation score to it. The company is adding ImagePrinting technology to this service that creates fingerprints specifically for images.

While spammers can circumvent fingerprinting by changing even one pixel in the image, Judge says Secure Computing's ImagePrinting technology performs image normalisation "to ignore variations and focus on reoccurring parts."

Tumbleweed last week introduced its Adaptive Image Filtering technology designed to block image spam by using a technique called wavelet transform, which reduces an image to a mathematical formula that represents the message but still allows for variation, the company says. With this new filtering technology added to Tumbleweed's e-mail security appliances and software, the products can catch image spam that has been randomised to circumvent spam filters, they say.

Whether or not anti-spam products can catch this new variant of spam, the huge increase in spam is concerning because it necessitates more bandwidth and computing power for anyone running an e-mail system, Levine says.

"Spam is a huge tax on e-mail and the tax just doubled," he says.