Gmail’s current machine learning model, the company claims, helps in blocking 99.9% of threats from reaching Gmail
inboxes and in order to make it more accurate, Google has added a new generation of document scanners that rely on deep learning to improve the company’s detection capabilities. This document scanner was added to Gmail at the end of 2019 and has led to 10% increase in the detection of malicious scripts in the Office documents on a daily basis. “ Our technology is especially helpful at detecting adversarial, bursty attacks. In these cases, our new scanner has improved our detection rate by 150%,” said Google.
Giving a bit more detail, Google said that this scanner uses TensorFlow deep-learning model and a customised document analyser for each file type. These analysers are responsible for parsing the document, identifying common attack patterns, extracting macros, deobfuscating content, and performing feature extraction.
Revealing some stats, Google said that malicious documents represent 58% of the malware targeting Gmail users. Google says that its new scanner runs in parallel with existing detection capabilities leading to better detection of malicious documents.
Earlier this week, Google announced new changes for both G Suite customers and users with personal Google Accounts alike.
As part of the new features, Google will be enhancing multiple inboxes to make them support independent scrolling, a unified toolbar and adjustable width in right-side configurations. After February 20, you will see a notification in Gmail informing you of this upcoming change if you currently have multiple inboxes in an unsupported configuration. Once this change takes effect, the preview pane will turn off if you use multiple inboxes.