There is an interesting article on viruses and filters in a Swedish newspaper today. Experts warn us that some filters are not efficient enough. They claim that some filters only catch 30% of the harmful e-mails. The reason for this being that the virus makers are quick at making new version and varieties of the code so it is hard to keep a traditional filter updated.
At SpamDrain we use artificial intelligence to continually refine the filtering process. This means that the filter can ‘learn’ what is spam and what is valid e-mail. The filter can therefore keep itself updated and quickly learn new tricks used by hackers to get their e-mails through spam filters.
This works quite nicely in Spamdrain when filtering spam and e-mail viruses. It would though be interesting to know if one could do something similar with desktop virus protection. I know that anti-virus companies have done some efforts in trying to “learn” the patterns of viruses and block viruses they don’t know about, but it’s unclear how well these techniques work.