Continuing its efforts to bring AI into every little product it has, Google announced today that improvements it made to Gmail’s machine learning algorithms help it detect and block spam and phishing messages with an accuracy of 99.9%.
To put things into perspective, the company says spam messages account between 50% and 70% percent of emails Gmail users receive on a daily basis.
If let’s say Gmail’s security systems would go down for a day, you can only imagine the amount of spam a user’s inbox would receive.
Blame it on Gmail’s new AI systems
Google credits this huge improvement to new security systems that rely on machine learning to analyze and catalog emails as they arrive or leave a user’s Gmail inbox.
In a series of four blog posts, Google also detailed a bunch of new security features it will be adding to Gmail in the following days. Here’s a summary of all updates:
Gmail will show a big fat warning inside the “Reply box” when a user is answering a user that’s not part of his organization’s domain, or that’s not in the user’s contacts list. This includes the reply-to address, but also CC and BCC fields. Google introduced this feature to prevent accidental replies to spear-phishing emails that arrive from unknown persons.
⋙ Gmail will add a delay of a few minutes to emails that match a known phishing pattern. Google says the delay will be up to 4 minutes, which is the amount of time its Safe Browsing technology needs to scan links included in the email.
⋙ Google says it now correlates spam signals with attachment and sender heuristics to predict messages containing new and unseen malware variants. Google says these improvements allow Gmail’s AI system to better detect zero-day threats, ransomware, and polymorphic malware.
⋙ Gmail on Android will receive the same click-time warnings available for desktop users. These are intermediary screens that Google shows when it thinks users have clicked on a phishing link it didn’t detect when the email was first received.