Mobile account takeover: when device ID isn’t enough
Protect your transactions! Protect your login! Protect your mobile channel! Protect your end users! A layered security architecture is now standard for most organisations. The problem, however, with many of today’s layered security solutions is that they do not correlate information between the various layers (security risks, suspicious events, fraud indicators etc.) and thus fail to see the big picture, writes Etay Maor.
Let’s consider a real world scenario: if you report a lost credit card, your credit card company knows to immediately suspend (or at least investigate) any card usage. The fact that the card was lost is correlated to future use. When it comes to malware, things are not that simple. Most end users don’t know their credentials were stolen by malware and therefore cannot report the compromise. This is also where security silos fail – an AV/anti-malware/anti-fraud solution may be aware that credentials are stolen but does not report crucial information to the criminal’s next stop – the login authentication system.
We recently discovered several fraud schemes targeting a large European bank. Among the interesting findings across factors such as device usage, fraud behaviour profiling and account access anomalies, one fact clearly stood out: there is a sharp increase in the use of mobile devices to conduct account takeover fraud. Even more interesting, the bank experienced fraud that originated from their mobile channel.
Why are criminals finding the mobile channel so attractive?
Let’s take a step back and examine things from the criminal’s point of view. One of the most popular fraud schemes involves stealing credentials (using malware and Phishing) to perform an account takeover (ATO). A recent Javelin Strategy report states that “Identity fraud incidence increased in 2012 for the second consecutive year, affecting 5.26% of US adults. This increase was driven by dramatic jumps in the two most severe fraud types, new account fraud (NAF) and account takeover fraud (ATF).”
Although stealing credentials is relatively easy (many fraudsters sell stolen credentials in the cyber underground forums), actually passing authentication procedures can be quite challenging. One of the most basic authentication methods is device ID. A criminal logging in from a new device is going to trigger a fraud alert, resulting in limited account access or even a failed login. If only there was a device that was both anonymous and yet unsuspicious to the targeted bank …
Mobile devices, and iPhones specifically, have an interesting trait – they all look the same. Not just physically, but also their device fingerprint. When a user browses to a web site from his native mobile browser (let’s take an iPhone and a Safari browser as our example), the device characteristics are identical to almost all other iPhones: same hardware, same browser, same fonts, etc. This scenario is a criminal’s dream come true and the reason why the bank discussed above was targeted from the mobile channel.
In this attack scheme, criminals use phishing and malware to steal credentials from the victims PCs. They then login to the bank using a mobile device and a native mobile browser (no mobile banking app is used). The bank cannot uniquely identify the device because the criminal’s iPhone looks exactly like the victim’s iPhone (or like any other iPhone for that matter). The criminal’s login attempt will not trigger any risk indicators and a fraudulent transaction is just a matter of time. This is exactly where security silos fail.
To address these types of threats, a mobile-aware risk engine must correlate fraud indicators across channels and over time. If this type of risk engine had visibility to the fact that the victim’s PC was infected with malware (with a client or client-less detection solution) and that shortly after the credentials associated with that account were used from a mobile device – fraud would have been stopped.
Marine Corps General Shoup said, “to lack intelligence is to be in the ring blindfolded”. If fraud prevention solutions don’t correlate risk factors to conclusively identify fraud, one cannot expect to knock out fraud.