Exploit code hiding in cache servers

Malicious code is living on weeks after it has been removed from websites thanks to an unexpected culprit - cache servers. According to Finjan Software, which has just released its latest Web trends report, caching technology used by search engines, ISPs and large companies has been discovered to harbour certain kinds of malicious code even after the website that hosted it has been taken down.

Such "infection-by-proxy" code can remain in caches for as long as two weeks, giving it a "life after death" at a time it would conventionally be assumed to have been neutralised. Although caching does not always save copies of everything on a website, it will still store code embedded in html, including programming formats such as Javascript.

The company offered details of how code designed to exploit a number of vulnerabilities in Microsoft products from 2003 and 2004 was able to continue in the public domain thanks to it hiding in the cache servers of one of three unnamed search engines.