We are banning more violent posts than ever, admits Facebook
FACEBOOK is becoming more violent, the social media site has admitted, as it revealed a sharp rise in banned content since the start of the year.
The social network, which has 40million British users, has released figures for the first time showing the levels of banned and blocked posts, from spam to extremism, on the site.
Violent material discovered on Facebook grew 183 per cent, with material from the war in Syria adding to the number of graphic images.
In total, 3.4 million violent posts and images were removed or blocked in the first three months of this year. Some graphic content was allowed to remain, but was put behind a warning message.
The rise represents a trebling of the 1.2million posts of violent material in the final three months of 2017. The number of terrorist posts increased dramatically, from 1.1million at the end of 2017 to 1.9million. Facebook said it took action to remove or hide 2.5million hate speeches, and 21million posts showing nudity or pornography.
The figures are part of a transparency drive at Facebook, which had come under pressure to clean up its act and block violent and extremist material.
The site has responded by employing thousands of moderators to assist its artificial intelligence software, although Facebook said artificial intelligence was getting better at identifying violent images and articles that broke the rules.
It said the rise in the amount of violence it found was due to improvements in detection tools, while it also fixed a bug that had made it miss some inappropriate content.
Facebook told The Daily Telegraph that the rise corresponded with a general rise in violence around the world, with images of war, from users and in news reports, widely shared.
“Terror attacks and incidents, like those in Syria, are allowed in a reporting context,” said Richard Allan, Facebook’s European public policy head. “Graphic violence is not always all bad content, but it could be shocking.”
Facebook said it was becoming more effective at banning extremist propaganda, its systems blocking more than 99 per cent of terrorist material promoted by the likes of Isil and al-qaeda.
Spam alerts also improved, with 837million posts removed before they went live, but it hasn’t stopped fraudulent adverts, such as those promoting Bitcoin, from tricking the system.
“We have a lot of work still to do,” said Alex Schultz, the data head. Youtube, owned by Google, has produced similar reports for several years. It said it had taken down eight million videos in the final three months of last year.