In addition to blocking malicious attachment types in emails, Google says Gmail will also now do a better job of protecting users from zero-day threats, ransomware, and polymorphic malware. To do this, Google is now taking spam signals and correlating them with attachments and sender heuristics. What Google says this does is assist in the prediction of malware and other threats that may have not yet been recognized by software and security experts, thus allowing users to be a little less worried about future malicious attacks through Gmail emails.
Google says between its implementation of machine learning within Gmail and the new security features that it's added to the email service as of today, that it expects to see a huge increase in the amount of additional emails that they can reject on a daily basis, stating that they would be able to reject hundreds of millions more emails per day, which ultimately reduces the number of threats that Gmail users will have to be subjected to. Of course it isn't all about security with Gmail as Google has done its fair share to improve Gmail in plenty of other ways too, like introducing other new features to the app and service, such as the recently added Smart Reply that Google started adding to the Gmail application as the middle of May. Today's inclusions to Gmail on the security side are just the latest security features to come with the service, and they certainly won't be the last as Google continues to improve the security of Gmail for its users.