You should soon receive fewer unwanted messages in Gmail: Google is starting to deploy a brand new antispam tool in its messaging which promises better efficiency in detecting intruders.

You should soon receive fewer unwanted messages in Gmail Google

You should soon receive fewer unwanted messages in Gmail: Google is starting to deploy a brand new antispam tool in its messaging which promises better efficiency in detecting intruders.

Even though ISPs and email services try to maximize your security, you still receive dozens of unwanted emails every day. A real plague that Google has tried to tackle. According to the American giant, its existing technologies already make it possible to delete 99.9% of spam before it even reaches your Gmail box. Despite everything, the Mountain View firm does not intend to stop there, aiming for 100%. To do this, Google recently developed a new device that could make life difficult for unwanted messages. This technology would allow even more spam to be filtered using artificial intelligence.

RETVec: a new new antispam for Gmail

For a long time now, cybercriminals have played with the limits of spam filtering systems. To ensure that their fraudulent emails slip through the cracks, hackers use homoglyphs, which are characters that resemble letters, keywords or even emojis. If these symbols are easy to spot for a human being, it is much more difficult for machines to realize the deception. From now on, however, these emails should become increasingly rare thanks to RETVec (Resilient & Efficient Text Vectorizer), Google’s new tool for filtering spam. As the Mountain View firm reveals in a statement published on its Security Blog, RETVec, which is notably based on a TensorFlow machine learning model, was deployed on Gmail.

Using artificial intelligence, the tool is able to analyze visual similarity to identify the meaning of words rather than their actual content. During internal testing, the company “notice that [RETVec] was very effective for security and anti-abuse applications”. Indeed, the results seem particularly promising. “Replacing the previous text vector of the Gmail spam classifier with RETVec allowed us to improve the spam detection rate by 38% compared to the baseline and reduce the false positive rate by 19.4%”explains the IT giant.

© Google

Besides the security aspect, RETVec has other advantages. For example, it allows “significantly reduce computing costs”but also to take charge “all languages ​​and all UTF-8 characters without requiring text preprocessing”. With a model with nearly 200,000 parameters, compared to millions for other tools, RETVec is lightweight enough to run on a local device without compromising performance.

Finally, Google’s new anti-spam is also available as open source on GitHub. This can allow developers to integrate the tool into their applications or research when needed. Besides, Google has thought of everything and published a tutorial to use it.

ccn3