Google Play Store ML detected almost 1 million violators last year

Android boasts of a more open and more welcoming ecosystem compared to iOS and part of that is its less stringent process in screening apps entering the Google Play Store. Creating a sort of chicken and egg situation, it meant that it would be logistically impossible for Google to manually review each and every app submission that comes its way. Since the beginning, the tech giant has relied on AI and machine learning to detect policy violators and now Google is boasting just how successful that technology has been in protecting Android users last year.

Google uses a mostly automated process when enforcing the security and integrity of the Google Play Store, both in screening apps before they get added as well as scanning apps already on the app store. Last year, it says that Google Play Protect scanned over 100 billion installed apps daily across billions of devices, hinting at how widespread Google's mobile empire has really become. The company also said that its machine learning enhanced app review process was able to detect over 962,000 policy-violating apps and banned around 119,000 malicious and spammy developer accounts.

Those are definitely impressive numbers given their size but security reports about malware on Android also paint a different picture. Quite a number do get through the cracks and are able to cause harm to affected users, from spamming them with ads to even making unauthorized purchases. Google is able to remove such apps but often only after they are reported.

That said, some types of apps do go through manual reviews and stricter requirements due to the sensitivity of their nature. Last year saw a surge in COVID-19 apps as well as election and news-related apps because of current events. Google created and implemented new policies just for these classes of apps in order to protect users' privacy and safety.

Google's efforts in securing Google Play Store and Android are definitely commendable, especially if it manages to really figure out the perfect automated system. Unfortunately, there are still a lot of cases to be covered and its machine learning system needs to learn more tricks to be able to block those malicious apps before they even set foot inside the Play Store.