Gmail introduces AI-powered spam detection to protect users from unwanted spam messages. The company wants to outsmart fraudsters with a new text classification system called RETVec.

Our inboxes are often bombarded with spam messages. The world-famous email platform Gmail wants to put an end to this and is relying on AI to do so. The new AI-based text classification system RETVec (Resilient & Efficient Text Vectorizer) developed by Google is a major step forward in the fight against text manipulation.

Debunking homoglyphs

Finally, text manipulation is a common tactic used by fraudsters. For example, by using homoglyphs or visually similar characters, fraudsters can imitate domain names. Such homoglyphs cannot be detected using traditional recognition technology. The new RETVec system, based on machine learning, focuses on visual similarity rather than mere character recognition, enabling homoglyph recognition. This new technology allows Gmail to better understand and combat text manipulation with advanced spam filters.

A long-term fight

Gmail has been actively optimizing spam filters for several years. For example, in 2019, the company used TensorFlow to detect 100 million additional spam emails per day. With RETVec, Google hopes to set a new benchmark in email security. Currently, the introduction of RETVec has improved spam detection by 38 percent false alarm reduced by almost 20 percent.