Google Boosts Detection of Malicious Documents in Gmail
New scanning capabilities that Google rolled out to Gmail have resulted in an increased overall detection rate of malicious documents.
The company says its existing machine learning models, combined with other protections, help block more than 99.9% of threats from reaching Gmail inboxes, and that its malware scanner processes more than 300 billion attachments each week.
Of the detected malicious documents, 63% differ from day to day, and the Internet search giant has deployed a new generation of document scanners to improve its detection capabilities via deep learning.
Since the end of 2019, the new scanners have already helped increase the daily detection of Office documents containing malicious scripts by 10%, Google says.
The company explains that the technology is particularly helpful when it comes to the detection of adversarial, burst attacks. In this regard, the scanner has improved the detection rate by 150%.
The new scanner, Google explains, uses a distinct TensorFlow deep-learning model trained with TFX (TensorFlow Extended), along with a custom document analyzer for each file type. These analyzers parse the docs, seek common attack patterns, extract macros, de-obfuscate content, and extract features.
With malicious documents representing 58% of the malware targeting Gmail users, Google is focused on further improving its document detection capabilities. Still in development, the new technology is used to scan only Office files at the moment.
“Our new scanner runs in parallel with existing detection capabilities, all of which contribute to the final verdict of our decision engine to block a malicious document. Combining different scanners is one of the cornerstones of our defense-in-depth approach to help protect users and ensure our detection system is resilient to adversarial attacks,” Google concludes.
The company is sharing the details of this technology at the RSA Conference.