The big security issue with social networks is the content posted by members. A new trend has emerged, where bad guys automatically create a large number of accounts via botnets, then grow their networks, and finally post links that are extremely dangerous to click on, reaching a large number of users. Sometimes the dangerous URL's are propagated via email, to all friends or connections, using the mailing client available on the network.
On social networks that encrypt the URL (such as Twitter), unless anti-fraud and link scoring algorithms are being used, the average user has no idea that clicking on a little tiny URL can result in very severe infections by viruses such as Gumblar, Conficker or Reader_s.exe. To compound the issue, the problems might happen more than 6 months after the initial infection, when the virus neutralizes the victim's computer after having stolen all the data including login and passwords.
Sometimes these tiny URL's generate a chain of redirects before landing on the infected server, making fraud detection even more difficult. Social networks not implementing the right anti-fraud technologies create significant risks for their users. On AnalyticBridge, we have implemented solutions to eliminate this problem, using proprietary URL scoring technology.