Facebook has announced that it is yielding the use of artificial intelligence systems and other automated techniques to remove terrorist-related posts, but there is still room for improvement.
Now, 99% of Al-Qaida and ISIS related material is detected automatically and not by other users. Facebook said that after the suspicious material was detected, 83% (and any copy of the same content) was automatically removed within just one hour of its posting. In fact, in some cases, posts were deleted “in their native” before they even became available online on Facebook.
But Facebook, according to the BBC and the Reuters agency, has recognized that there is still a lot to be done to detect other terrorist and extremist groups that distribute material through the largest social media tool.
While until recently, Facebook used a mixture of people and software to identify which posts should be removed, this work is almost exclusively done by automated systems that analyze texts, photos and videos that have their own separate “digital footprints” .
Artificial intelligence software with mechanical learning is “trained” to gradually recognize the content of texts related to terrorism, based on specific words and phrases.
The company said it has so far focused its efforts on Al-Qaida and ISIS (Islamic State), because they were the biggest threat worldwide. But assured that is now expanding her net to other groups, though, said, identifying the posts of all these different teams, “it will not be as easy as pushing a button.”
The US government and several European leaders have said they would like Facebook with its 2.1 billion users to move faster so that all terrorist and extremist material is deleted within two hours of its posting. European governments have let the threat of fines against Facebook hang if it does not respond.
Next week, Facebook, Twitter, Alphabet and other technology companies will meet with EU governments and the European Commission to discuss how to remove and accelerate the removal of terrorist and extremist content from the media of social network.