Thousands of sites loaded with potent malware cocktail
'Ugly can of worms'
Agentless Backup is Not a Myth
Cybercriminals have laced about 2,000 legitimate websites with a potent malware cocktail that surreptitiously attacks people who browse to them, a security researcher warned Friday.
Unlike past outbreaks of the mass web attack known as Gumblar, this round actually plants exploit code on the website servers themselves. Curiously, the directory and file name of the malicious payload is in most cases unique and identical to a legitimate file that existed on the website.
The trick makes it extremely difficult for webmasters and anti-malware programs to detect the threats.
"This is an ugly can of worms," said Mary Landesman, the ScanSafe security researcher who warned of the mass attack. "Any time you see a new technique evolve like this the concern is we'll be seeing much more of this in the future, and certainly it complicates the remediation of the compromised website."
Previously, Gumblar planted links in thousands of compromised websites that silently redirected users to a handful of servers that hosted the exploits. That method allowed white hats to foil the attack by shutting down one or two domains. With the malware embedded directly in the compromised websites, the take-down process is significantly more time consuming.
Also making matters hard for Landesman to get the sites cleaned up: Most of the websites belong to small businesses that cater to non-English speakers. Few of them have dedicated security employees, and even when representatives can be located, the person contacting them must speak multiple languages.
While the websites are relatively small, Gumblar architects have planted links in online discussion forums across the web that often cause RSS readers to automatically send users to the booby-trapped pages. Landesman suspects black-hat search engine optimization may also be causing the infected sites to be featured prominently in results returned by Google and others.
People who are unfortunate enough to visit the sites won't see anything unusual. But behind the scenes, a PHP script checks their version of Adobe Reader and Adobe Flash, and if either is out of date, hijacks their PCs using known vulnerabilities. If both of those programs are up to date, the script tests to see if the system is vulnerable to several bugs Microsoft has patched in the last few months.
Hijacked machines will be installed with a backdoor that gives the hijackers complete control. They are also equipped with malware that manipulates search results returned by Google.
It's unclear exactly how the sites are getting compromised. Landesman suspects FTP passwords for the sites have been lifted from administrators' computers using key-logging malware. ®
COMMENTS
@Sitaram Chamarty
Yes, you DO have to take them out before you shoot them -- otherwise you get blood on the carpet. ;^P
server side script
The script is server-side; it's PHP which means it'll happily reside on *nix servers... and in all probability it's (cheap) *nix hosts that have been targeted. That does not mean it will affect *nix clients however - it simply uses PHP to glean some information about the user's system, looking for known vulnerabilities - as the story states - initially in Adobe software and then MS vulns.
Unless you're doing something seriously daft, you should be safe under *nix (depending on payload). Under windows, noscript _may_ prevent the attack if it happens to be blocking Flash at the time of viewing a compromised site. If. however, it's a site that you've already white-listed, all bets are off... and that's assuming you've got Windows itself patched up to date.
My money is on an FTP breach - there have been a few in recent months, primarily targeting the cheap *nix hosting market - it seems to be something in the way the hosting companies have their systems set-up (open or anonymous FTP access - no IP address restrictions) - using keyloggers was the first "explanation" uttered (e.g. it's not our fault, it's yours for not securing your PC) but it could be packet sniffing or brute force... whatever.
This recent spate of cracks have normally resulted in .htaccess uploads (full of Mod Rewrite redirects) - it was only a matter of time before someone combined something genuinely dangerous with these breaches.
Easy answer
Make the owner of the vulnerability responsible for repair, if they leave people's machines open to attack they should have to pay to repair any damage done by their bad coding

IT infrastructure monitoring strategies
Requirements Checklist for Choosing a Cloud Backup and Recovery Service Provider
Data control in the cloud
Cloud based data management
Agentless Backup is Not a Myth