Facebook purges millions of posts in just 4 weeks, says AI acted even before users reported
A Facebook spokesperson said that the social networking firm would continue to use a combination of Artificial Intelligence, community reports and reviews by its teams to identify and review content against the platform's policies.
FollowFacebook Inc has published the first edition of its compliance report under the new Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, which revealed that the social networking giant had taken action upon 30 million content pieces across ten violation categories during May 15-June 15.
As per the new intermediary guidelines that came into force on May 26, all significant social media intermediaries (over 50 lakh users in India) had to publish monthly reports mentioning the details of complaints received and the action taken on them. The firms are expected also give the number of specific communication links that the platform removed through proactive monitoring.
The Facebook monthly report comes days after Google released its data, in which it disclosed that it had removed 59,350 pieces of content.
Here are the key takeaways from the Facebook report:
* Facebook took action against 25 million spam content (99.9% proactive rate).
* Facebook took action against 2.5 million violent and graphic content (99.9% proactive rate).
* Facebook cracked down on 1.8 million adult nudity and sexual activity-related content (99.6% proactive rate).
* Facebook also removed 53,000 pieces of hate speech.
* Facebook's proactive monitoring rate for harassment and bullying on the platform remained low at nearly 37%.
* Facebook-owned Instagram removed 6.99 lakh posts linked to suicide and self-injury content (99.8% proactive rate).
* Instagram also removed 490,000 posts linked to adult nudity and sexual activity.
* Over 90% of the content has been removed through Facebook's Artificial Intelligence system before any user reported it.
* Instagram's proactive monitoring rate for bullying and harassment content remained low at 43% per cent.
A Facebook spokesperson said that the social networking firm would continue to use a combination of Artificial Intelligence, community reports and reviews by its teams to identify and review content against the platform's policies.
Facebook noted in its report that the rate at which it can 'proactively detect potentially violating content is high for some violations'.
"We find and flag most content before users do. We have been able to build machine learning technology that automatically identifies content that might violate our standards," Facebook said in its report.
