Security researchers from Indiana University, Bloomington, who recently disclosed multiple serious security flaws in iOS and OS X platforms have discovered what they say is an alarmingly high number of potentially harmful Android applications in Google's Play store.
More than 7.5 percent of the Android applications reviewed by the researchers turned out to be potentially harmful: the apps collected user’s private data, installed app list, locations, contact lists, and photos -- all without the user’s consent.
Some 400 of the apps have already been downloaded over one million times each, while an additional 2,000 Android apps were downloaded over 50,000 times.
Other Android application stores that the researchers inspected including Amazon Appstore, Samsung Galaxy Apps, and tens of smaller third-party markets, did not fare much better, according to Kai Chen, one of the researchers involved in the study.
For example, 7.8 percent of the Android apps on Opera, a popular European market for mobile apps, were found to be potentially harmful, while 5.9 percent of the apps on Amazon’s Appstore fell into that same category, Chen says. At least 20 of the malware samples Chen and his team unearthed across all app stores were potentially zero-day threats.
Chen and his counterparts used a malware scanning technique that they developed themselves called MassVet (short for "mass vetting") to find the potentially malicious applications on the different app stores. The scanner exploits the way most Android malware is constructed and distributed, Chen says.
In most cases, purveyors of Android malware simply repackage legitimate applications with malicious components and slip them into application stores, where unwitting mobile users download them.
"This practice makes such malware stand out from other repackaged apps, which typically incorporate nothing but advertising libraries," Chen and his counterparts wrote in a 16-page report technical documenting their research and how MassVet works. The "Finding Unknown Malice in 10 Seconds: Mass Vetting for New Threats at the Google-Play Scale" report is available for download here.
Usually, the same malicious components are used across multiple applications. "As a result, such attack payloads often stand out from those of the same repackaging origin and also show up in the apps not supposed to relate to each other," Chen wrote.
By comparing a potentially harmful app to the legitimate one it impersonates, it becomes easy to spot the difference in code, he says.
"Unlike existing detection mechanisms, which often utilize heavyweight program analysis techniques, our approach simply compares a submitted app with all those already on a market," the researchers wrote. "Focusing on the difference between those sharing a similar UI structure (indicating a possible repackaging relation), and the commonality among those seemingly unrelated," makes it relatively straightforward to spot potentially harmful applications.
Once public libraries and other legitimate code reuse are removed from an app, any different components that might remain become highly suspicious.
The researchers ran MassVet against a total of 1.2 million applications that they collected from 33 Android markets. About 400,000 of the apps were from Google Play, 596,437 from 28 app stores in China, some 61,850 from European stores, and 27,047 from what the researchers described as "other US stores."
From the 1.2 million apps scanned, MassVet weeded out 127,429, or slightly more than 10 percent as being malware, using the app–to-app comparison approach. "Among them at least 20 are likely zero-day and 34,026 were missed by the majority of the malware scanners run by VirusTotal," they wrote.