Using these techniques, attackers are now able to rapidly deploy thousands of unique, malware laden messages that are largely undetectable to traditional signature and reputation-based security systems. Worse, despite their scale, these mass customized phish were effective enough to trick more than 10 percent of recipients into clicking on malicious content capable of taking complete control of PCs and compromising corporate networks.
Phishing meets mass customization
Unlike conventional mass phishing exploits, the "hooks" (email messages) used in longlining are highly variable rather than identical, making them largely undetectable to traditional signature and reputation-based security gateways.
The messages are typically varied by IP address of origination, subject line and body content. The body content also includes multiple mutations of an embedded destination URL, which typically leads to a site with a positive reputation thatís been successfully compromised prior to the attack. The compromised Web destinations are loaded with hidden malware either before, during or sometimes after the attack wave has begun.
Through the use of a distributed cloud of previously compromised machines and process automation to create high variance, attackers have been able to combine the stealth techniques and malicious payloads of spear phishing with massively parallel delivery. This means they can cost-effectively send 10,000 or even 100,000 individual spear phishing messages, all capable of bypassing traditional security. Attackers' ability to distribute thousands of email-borne malicious URL "hooks" in a matter of hours greatly improves their odds of success and their ability to exploit zero-day defects before corporate IT has time to patch vulnerable systems.
As part of the six month study, which involved over one billion email messages, Proofpoint observed, documented and countered dozens of longlining attacks globally.
For example, on October 3, 2012, the company observed a Russia-based attack with 135,000 emails sent to more than 80 companies in a three-hour period. To avoid detection, the attacker employed approximately 28,000 different IP addresses as sending agents, 35,000 different "sender" aliases, and more than twenty legitimate websites compromised to host drive-by downloads of zero-day malware.
Because of the different agents, sender aliases, URLs and text, no single targeted organization saw more than three emails with the same characteristic. Overall, this attack represented less than 0.06 percent of the targeted companiesí mail flow (compared to 19 percent for spam and 11 percent for virus-laden email). The combination of mass customization and proportionally low volume made this longlining attack effectively invisible to traditional anti-spam products, enabling widespread access to corporate networks.
Similar attacks were documented throughout the fourth quarter of 2012 and early 2013. In another representative attack, approximately 28, 800 messages were sent in multiple one-hour bursts to over 200 enterprises. The campaign consisted of 813 unique compromised URLs sent from 2,181 different sending IPs. Again, each organization saw no more than three messages with identical content.
Despite their relatively large scale, longline attacks were alarmingly effective:
- Ten percent of the email messages containing embedded malicious URLs that escaped perimeter detection were clicked on by the receiving employees
- All the longline attacks employed "drive-by downloading" from compromised web-sites.
- Almost one out of every five clicks (19%) on malicious URLs embedded in email occurred "off network" when employees accessed their email from home, on the road, or via mobile devices where they were outside corporate perimeter protection.