It's not just the Recording Industry Association of America (RIAA) that people need to worry about when downloading music from peer-to-peer (P2P) networks.

A surprisingly high number of consumers sharing music and other files on P2P systems are inadvertently exposing all sorts of bank account and similar personal information on their computers to criminals lurking on the networks to harvest data. And its not just users at home that are exposing information about themselves; so are a large number of employees within banks, as well as their contractors and suppliers.

That's the conclusion of a study on the dangers of inadvertent data disclosure on file-sharing networks that was conducted by Dartmouth University's Tuck School of Business.

The study examined data involving P2P searches and files related to the top 30 US banks over a seven-week period between December 2006 and February 2007. The university used a search engine technology from Triversa to gather and analyse all P2P traffic that mentioned these banks by name or mapped to a specific digital footprint that Dartmouth created for each financial institution. Data was gathered from P2P networks such as Gnutella, FastTrack, e-donkey and Bittorrent.

The analysis showed that a large number of searches made on those networks were aimed at uncovering sensitive financial data from individuals, said study author Eric Johnson, a professor of operations management at the school's Center for Digital Strategies. "Our analysis clearly reveals a significant information risk firms and individuals face from P2P file-sharing networks," he said.

When people use popular P2P clients such as KaZaa, Limewire, BearShare, Morpheus and FastTrack, they often are sharing far more than just media files, Johnson said. "In many cases they are sharing the contents of their entire hard drive with all sorts of information" with others on the file-sharing network, Johnson said.

That's because many of these client tools are designed specifically to quickly search for and share certain types of media files on a user's system. Normally, such P2P clients allow users to download files to and share items from a particular folder. But if proper care is not taken to control the access that these clients have on a system, it is very easy to expose far more data than intended, he said.

There are several ways this can happen, Johnson noted in his research paper. For instance, when a music file is accidentally dropped into a folder containing other data, the contents of the entire folder could end up being shared on a P2P network without a user's knowledge. Many P2P client software tools have confusing interfaces that could result in users sharing folders that they did not intend to. Similarly, some file-sharing apps feature wizards that scan an individual's computer and recommend folders containing media to share. If a sensitive file exists in one of those recommended folders, it could get exposed, Johnson wrote in his research.

The kind of information that can be exposed in this manner is astounding, Johnson said. "We found files containing all the information needed to commit identity theft. We found almost every kind of business documents from spreadsheets to performance reviews. In one instance, we found a bank spreadsheet with account information on 23,000 business accounts that was leaked. We even found a security evaluation done by a third party contractor" of a bank network.

Almost 80 percent of the leaked information analysed in the Dartmouth study came from home PC users. The rest came from systems belonging to bank employees or their partners, Johnson said.

While some of the information was inadvertently leaked, there are growing signs that cybercriminals are using P2P networks to specifically search for and harvest such data, Johnson said. A significant portion of the search terms that were analysed during the Dartmouth study appeared to be looking for databases, account and user information, passwords as well as routing and pin numbers, Johnson said, Sometimes, sensitive data was accidentally exposed by the coincidental association of a search term with sensitive information. For example, users searching for songs containing the terms ‘Golden’ Or ‘West’ in the title pulled up files containing account information belonging to Golden West bank, Johnson said in his report. Similarly users looking to download the song ‘State Street Residential’ sometimes pulled in data belonging to State Street bank customers.

The Dartmouth study raises concerns similar to those contained in a report released in March by the US Patent and Trademark Office (USPTO). That report was based on an analysis of five specific features included in file-sharing software from Kazaa, Limeware, Morpheus, BearShare and eDonkey. The report concluded that the distributors of the software deliberately included these features in their tools, despite knowing that the features could cause users to inadvertently share sensitive data with others on P2P networks.

The report was sent to the US Department of Justice, the Federal Trade Commission and the National Association of Attorneys General.