Facebook has revealed a new AI-powered tool to help it fight fake and spam accounts.
Known as Deep Entity Classification (DEC), this machine-learning tool can not only identify fake accounts, but is also able to analyse active accounts and study users behaviour within the Facebook community.
The company claims that it has already used DEC to take down over 6.6 billion fake accounts last year, as well as blocking millions of attempts to create new accounts.
- Dark mode lands in Facebook Lite – is the main Facebook app next?
- Facebook Pay brings new payments to Instagram, WhatsApp, and Facebook
- Facebook sues analytics firm over alleged data harvesting
Facebook artificial intelligence
Facebook added that DEC has helped them to reduce the amount of fake and scam accounts on its system by 27%.
The company believes that almost 5% of its 2.89 billion monthly accounts users are violating its policies, generally through spamming or using phishing scams to try and steal money or sensitive personal information.
DEC looks beyond general information such as account creation date or the number of friend requests sent to also look at more complex attributes like the pages that a user’s profile has liked, groups it has joined to try and identify the authenticity of a profile.
Facebook claims that DEC is able to analyze over 20,000 features per profile, and could well play a vital role in the upcoming presidential elections in the United States.
The company has already identified profiles from countries like Russia, Iran, Myanmar and Vietnam that were engaged in using the platform to influence voting behaviours and news narratives.
- We'll help you pick from the best VPN options - and the rest