Google eliminates more spam from Gmail with TensorFlow

Yolanda Curtis
February 8, 2019

TensorFlow, an open source software library that developers can use to build AI tools.

TensorFlow, which Google made open source back in 2015, has been used to improve spam detection for a while, but it's now received broader adoption and a larger set of spam classifications "at scale", said Neil Kumaran, product manager at Google's Counter Abuse Technology division, speaking to VentureBeat.

It an achievement it calls "a feat", Google claims it now blocks an additional 100 million spam emails every day. Google says that for Gmail, recognising spam and blocking has been a challenging task given the varied forms and it's constant evolving and incremental nature as well. The company stops more than 99.9 percent of spam, phishing, and malware from ending up in Gmail users' inboxes through the use of AI and other protections.

Gmail has around 1.5bn users, which means that 100m extra spam messages blocked is not as monumental as it sounds.

Google is now blocking spam categories that were previously very hard to detect.

Google added the new machine learning framework from TensorFlow does not replace the existing technologies that were already in place, but it simply exists alongside them. All in all, TensorFlow allows Gmail to scale its ML efforts, requiring fewer engineers to run more experiments and protect users more effectively.

What kind of additional spam messages did Google now get better at deflecting? Thanks to TensorFlow, Gmail is now able to block more complex messages, like image-based spam emails, emails with hidden embedded content, and emails from newly created domains that are used to hide low volume of spammy messages within legitimate traffic.

Google has become even more efficient at blocking spam email from reaching people's inbox, and it's praising its machine learning capabilities for its success.

Google has said that, by integrating TensorFlow into Gmail, it can better personalise spam filters and that this is the turning point for understanding spammers signals and "turning those signals into better results".

Other reports by iNewsToday