bad apps from its Play mobile app store last year.
The number represented a 70 percent increase over the number of apps that Google removed from the store in 2016.
Not only did Google remove more apps that violated Google Play policies, it also was able to identify them and boot them out sooner, Google Play product manager Andrew Ahn said in a blog Jan 30. “In fact, 99 percent of apps with abusive contents were identified and rejected before anyone could install them,” Ahn said.
He attributed Google’s better app removal performance last year to machine-learning enabled improvements in the company’s ability to detect malware, inappropriate content and spoofing in apps loaded to the Play store.
New detection models and techniques that Google deployed last year also improved the company’s ability to identify and remove repeat offenders and abusive developers at scale. In total, Google removed as many as 100,000 rogue developers from Play last year while making it harder for bad actors to register new accounts with Google, Ahn said.
The security of apps loaded to the Google Play store has become a big concern for the company. On numerous occasions over the past year and more, threat actors have managed to upload Android apps containing malware, spyware and other unwanted features to Google store. Many of the rogue applications were downloaded by tens of thousands of people—in some cases even over a million—before they were removed.
Google has over the years instituted numerous manual and automated controls for preventing, detecting and removing such apps from Play store. In 2017, Google launched Play Protect, a tool for continuously monitoring for and removing malware and other unwanted apps from Android apps running on end user devices.
Such measures have helped Google exert more control over the quality of Android apps that developers upload to Google Play. Even so, there have been numerous occasions in recent months where the company has completely missed spotting bad apps on Google Play and removed them only after being notified of their presence by third-party security vendors and researchers.
Now Google wants the public to know just how effective and comprehensive its app screening and removal process is.
According to Ahn, the types of bad apps that Google removed from Play last year included copycat apps that attempted to deceive users into downloading them by impersonating other, more popular apps. These represented more than 250,000 of the apps that Google removed last year.
Potentially harmful applications—such as those designed to serve up ads, steal a user’s personal information or conduct SMS fraud—continued to be another significant threat to Android users.
“While small in volume, PHAs pose a threat to Android users and we invest heavily in keeping them out of the Play Store,” Ahn said. Google’s new detection methods reduce the install rate of such apps on Google Play by 50 percent, he noted.
Applications containing inappropriate content were another big category. Google’s improved machine learning models helped the company more effectively spot and remove applications containing pornography, extreme violence and hate. Tens of thousands of applications that met Google’s criteria for inappropriate content were removed last year, Ahn said.