Google has announced that it recently added deep learning capabilities to its malware scanner for Gmail as part of an effort to detect and block malicious attachments. The new technology will be demonstrated at the RSA 2020 event in San Francisco, California.
The search giant's existing document scanner handles more than 300 billion attachments every week to block malicious content. Google notes that 63% of the malicious documents it blocks take different forms on a daily basis. To help address this increasingly evolving threat landscape, the deep learning-based document scanner will work to bolster Gmail's detection capabilities.
Gmail's malware scanner was launched late last year in order to ward off spam, phishing attempts, and malware from reaching your inbox. The technology currently relies on machine learning and Google claims it is, so far, effective at blocking more than 99.9% of threats from reaching Gmail inboxes.
Since its launch, the malware scanner has helped grow the daily detection coverage of Office documents infected with malicious scripts by 10%. With the addition of deep learning, Google claims its detection rate has improved by 150% as its new scanner uses a TensorFlow deep-learning model. The new document scanner is still in its infancy at the moment and Google says it is used only to scan Office documents for now.