Google Uses Machine Learning for New Security Features on Gmail

Use this guide to understand Google uses machine learning for new security features on gmail and the factors that affect it. The information is organized to make the topic easier to apply in practice.

Google is pushing four new security features for business users on G Suite, its business suite. These security measures were added immediately after Citizen Lab reported being attacked by fake emails from Russian hackers and received false information falsifying Google services, including Gmail.

How Google Uses Machine Learning for New Security Features on Gmail Works

Google does not mention Citizen Lab's report in the article introducing new security features, but mainly these methods are intended to help counter phishing techniques (phishing) to steal data and information. sign in.

The attack is aimed at more than 200 people in 39 countries. Among those attacked include "former Russian prime minister, European cabinet member, Asian continent, ambassadors, senior military figures, CEOs of energy companies and many social members. civil, "the report said.

One of the first emails of this attack that Citizen Lab followed was a fake Google security alert, sent to a Gmail account user. The malicious links are sent in the email, then redirect users to the fake Gmail page to sign in and thereby steal the login information. Redirecting the page makes the link look legitimate because the URL will display the Google domain.

Five days after the CItizen Lab report was released, Google announced it would release four new security features in the G Suite, including the ability to use machine learning to block spam and phishing. This method will select and delay a few emails (less than 0.05%) to perform fake analysis. This process is then supported by integrating Google Safe Browsing to detect suspicious URLs.

"This new model incorporates many techniques, such as analyzing reputation and the similarity of URLs, allowing for the creation of alerts when there are fake links or malicious code. When finding new trends, the model will Quickly adapt to old craft systems and improve themselves, "said Andy Wen, Product Manager at Counter Abuse Technology at Google.

Another form of security enhancement for DLP-based G Suite has an unintentional external page response warning. These alerts appears when users try to respond to someone outside the company's domain. It will self-exclude users outside of user or interactive contacts. Contaminated attachments are also a problem that the G Suite will update with the latest updates.

According to Sru Somanchi, Gmail Product Manager " We are looking for a link between thousands of spam mail alerts with attachments and senders, thereby predicting emails containing unprecedented malicious variants. ". This includes blocking high-risk file types such as JavaScript or executable files.

" Machine learning has helped Gmail achieve 90% more accuracy in detecting spam messages and with these new protection methods, we can minimize the chance of risk by blocking messages every day. ", Somanchi added. All of these updates will be available in the G Suite on weekends.