Spam messages could soon imitate a friend or colleague鈥檚 individual writing style, warn computer security experts.
Canadian researchers have developed software that generates 鈥減ersonalised鈥� spam messages by scanning legitimate email messages. They warn that real spammers may start using the same trick to create messages that slip past spam filters.
Large quantities of spam is already sent from personal computers that have been hijacked using computer viruses. These 鈥渮ombie鈥� machines can churn out millions of unsolicited advertising messages to addresses collected by web-trawling programs.
Advertisement
But John Aycock, from the University of Calgary, says spammers could also use these machines to create more convincing spam messages. 鈥淎ll the pieces are already there,鈥� he says. Aycock believes spammers will modify the software used to control zombie computers so that it also digs though sent email messages. This would allow it to create realistic emails containing spam links to friends, colleagues and others, he says.
Millions of messages
To prove such a scenario was possible, Aycock鈥檚 team developed software that analyses sent emails in order to produce spam. They used their own email messages, as well as the half a million emails disclosed as part of the Enron fraud investigation.
The software scans every sent email in a person鈥檚 mailbox, recording common misspellings, abbreviations, capitalised words and the style used to sign off. It also collects information about social relationships, by recording which people received the same messages, for example.
The program then uses this information to create more convincing spam messages. It can, for example, automatically reply to a real email with a message that includes the name of a mutual acquaintance as well as a spam link. The system has yet to be tested using a proper spam filter, but Aycock says the messages generated are more realistic.
鈥淚t鈥檚 really quite easy to extract this kind of information, and I think the emails the software generates are pretty convincing,鈥� he told 麻豆传媒. 鈥淚f a message looks like it came from a friend and it says 鈥榗heck out this cool link鈥� what are you going to do?鈥�
Loss of trust
Aycock says more sophisticated anti-spam tools will be needed to beat this type of spam. But he is unsure exactly how it could be done: 鈥淔or regular users this might mean we just have to be less trustful of email.鈥�
However Julian Field, an expert on spam filtering based at the University of Southampton, UK, does not foresee such a bleak future. 鈥淚t鈥檚 very clever but I鈥檓 not convinced,鈥� he told 麻豆传媒.
Field says checking that a message has come from a legitimate email client could catch such spam. 鈥淭he best way is to restrict only certain programs, email clients, to sending email,鈥� he says. This means that hidden programs, placed on a machine by spammers, could not send out email. Field adds that building this restriction into a computer operating system, such as Windows, would also warn users that their machine was sending out spam.
Aycock will present details of his research on 30 April at the annual conference of the European Institute for Computer Anti-Virus Research in Hamburg, Germany.
