Google is building the ultimate spam-killing AI

Google is using artificial neural networks to filter spam

Google engineers have cooked up a new way for Gmail spam filters to learn what mail you want and what you don't. It involves 'large scale brain simulations'.

In a post on the Gmail blog, product manager Sri Harsha Somanchi wrote that less than 0.1% of the email in an average Gmail inbox is spam, and the false positive rate is even lower - at just 0.05%. "Even still, Gmail spam detection isn't perfect," he said.

So the company is now applying artificial neural networks to detect and block the kind of spam that often passes for wanted mail.

Thanks to all the times you've clicked the "report spam" button, Google now has a pretty good idea about what you want and what you don't.

AI Know What You Want

Its AI will take those preferences and use them to customise your spam filter. "While your neighbor may love weekly email newsletters, you may loathe them," wrote Somanchi. "The spam filter can now reflect these individual preferences."

It's also adding new "Postmaster Tools" - aimed at companies that send a lot of emails. Any firms that meet Google's reputation requirements will be able to see how often its users mark their messages as spam - allowing them to tweak them so they're taken more seriously.

Duncan Geere
Duncan Geere is TechRadar's science writer. Every day he finds the most interesting science news and explains why you should care. You can read more of his stories here, and you can find him on Twitter under the handle @duncangeere.