Google used machine learning to remove over 700,000 malicious apps from its store in 2017

Google is working hard at stopping malware from sneaking into its Play Store to abuse billions of Android users. The number of malicious apps removed from the store rose more than 70 percent in 2017 from 2016 to 700,000, thanks to an improved machine learning detection algorithm for malicious and abusive techniques, according to the company’s end-of-year report.

Besides malicious programs, Google engineers also removed more than 100,000 developer accounts linked to cybercriminals abusing the store.

“In 2017, we took down more than 700,000 apps that violated the Google Play policies, 70% more than the apps taken down in 2016. Not only did we remove more bad apps, we were able to identify and action against them earlier,” writes Andrew Ahn, Product Manager, Google Play.

“In fact, 99% of apps with abusive contents were identified and rejected before anyone could install them. This was possible through significant improvements in our ability to detect abuse – such as impersonation, inappropriate content, or malware – through new machine learning models and techniques.”

The deleted applications fall under three categories, Ahn explains. The first is copycat applications impersonating popular titles. The second, and most pervasive category, represents applications with content related to pornography, hate and extreme violence. The third is dubbed Potentially Harmful Applications, meaning the apps are malicious and could lead to SMS fraud, turn into Trojans or phishing attempts.

“The annual PHA installs rates on Google Play was reduced by 50 percent year over year,” according to Ahn, whereas the second category is still extremely widespread. Although Google regularly improves its detection capabilities, the tech giant admits that some malicious apps could bypass security layers, so users should remain vigilant.