GOOGLE INTRODUCES A NEW AI-POWERED SPAM DETECTION TOOL FOR GMAIL

Fri 15 December 2023:

Google has brought up a new AI-powered feature to detect spam. This new feature will help in spotting ‘adversarial text manipulations’ like emails with special characters, typos, emojis and other characters which can easily bypass the Gmail defence.

Confirmed as ‘one of the largest defence upgrades in recent years’, this Google upgrade comes in the form of a new text classification system named Resilient and Efficient Text Vectorizer (RETVec).

In an official statement, Google wrote, “To help make text classifiers more robust and efficient, we’ve developed a novel, multilingual text vectorizer called RETVec that helps models achieve state-of-the-art classification performance and drastically reduces computational cost,” the company said.

Systems such as YouTube, Google Play and Gmail are relying on text classification models to identify harmful content which includes inappropriate comments, phishing attacks, and scams.

These kinds of texts are harder for machine learning models to classify because the bad actors rely on adversarial text manipulations to actively attempt to evade the classifiers.

Google further said, “For example, they will use homoglyphs, invisible characters, and keyword stuffing to bypass defenses.”

Due to its novel architecture, RETVec has been working out-of-the-box on every language and all the characters without any need for text preprocessing- making it an ideal candidate for web, on-device, and large-scale text classification deployments.

The tech giant said, “Models trained with RETVec exhibit faster inference speed due to its compact representation. Having smaller models reduces computational costs and decreases latency, which is critical for large-scale applications and on-device models.”

What is RETVec?

RETVec stands for Resilient and Efficient Text Vectorizer and it is a novel open-source text vectorizer which enables people to build more resilient and efficient server-side and on-device text classifiers.