Google is fighting malware on Gmail with machine learning

Microsoft Office documents are the main target for Gmail users.

Gareth Fuller - PA Images/PA Images/Getty Images

Email still reigns as one of the most popular places to target users and their data, mainly through the form of malicious attachments. Widely-used email clients such as Outlook, Apple Mail, and Gmail have been great at filtering out emails laced with these types of malware, but Google wants to do even better.

Google opts in for more security — In a blog post on Wednesday, Google announced its expansion with AI to better improve detection and detailed its process for spotting malicious documents. According to the company, 58 percent of all malware targeting Gmail users is spread in this way. The majority of these malicious files, Google says, are Microsoft Office documents.

Google has been working to boost its security efforts across all of its products, including Nest and Android. In October 2018, Google tweeted it reached over 1.5 billion active Gmail users — way too many people to leave vulnerable to security threats.

Machine learning vs. malware — Google’s malware scanner currently processes more than 300 billion attachments weekly and 63 percent of the documents scanned vary on a daily basis. The company says it currently blocks 99.9 percent of threats targeting Gmail users' inboxes.

At the end of 2019, the tech giant launched new scanners equipped with machine learning to ramp up its detection accuracy. Since then, Google states it's been able to increase detection of Office documents by 10 percent, improving the overall detection rate by 150 percent. The scanners work in tandem with all of Google’s existing anti-phishing and anti-scamming features.

And, Google says it still has its eyes set on even more improvements in the future.

“We will continue to actively expand the use of artificial intelligence to protect our users’ inboxes, and to stay ahead of attacks,” Google stated in a blog post.