Gnutella users warned of Internet worm
Novel virus spreading technique
Security firms are warning about a proof-of-concept Internet worm which tries to spread malicious files via the Gnutella peer-to-peer file sharing system.
The virus, called Gnutella Mandragore, will connect to the Gnutella network once it affects a users PC and will then attempt to offer itself for download when users search for songs or other files to download.
For example, if a Gnutella user makes a search for "Britney Spears swearing", the infected node will announce it has available a file called "Britney Spears swearing.exe", 8kB in size.
If a user downloads and clicks on this file, his machine becomes infected and will start to offer infected files for other users.
Infected nodes easily become overloaded and are unable to answer all requests. So every search does not find infected files although there are infected nodes in the network.
The first infected files in the Gnutella network were spotted last Friday but the worm is not spreading fast enough to become a serious risk.
An easy way to avoid infection from this worm is not to download exe files from the Gnutella network, and to be careful of files of 8kB in length. This is made easier because some Gnutella clients (such as BearShare) hide executable files from search results by default.
Discussion of how peer to peer networks might become a vector for spreading malicious code has been understood as a risk in security circles since last June, when the topic was discussed on security mailing list BugTraq.
Jack Clark, European product manager at Network Associates, said: "The more ways you share information the more ways there are to share malware. The benefits of opening up your hard drive to file sharing networks also carry risks which need to be protected against." ®
Napster alternative: hack people's hard drives
Gnutella leaves foolish people open to fraud
Microsoft preps Napster clone
Anna Kournikova bug drops harmlessly onto the Net
Britney gets filthy in front of 170,000 fans
Sponsored: Hyper-scale data management