Facebook announced that over the past two months, nearly 66,000 publications a week had been removed as a hate speech. The social network revealed this information on June 27 as part of a series of data aimed at elaborating its plan of action on how to deal with controversial topics such as fake news.
"We oppose hate speech in all its forms, and we do not allow it on our platform," said Richard Allen, vice president of public policy for Europe, the Middle East and Africa. "Our definition of a hate speech is anything that attacks people directly, whether in race, national origin, religious affiliation, sexual orientation, sex, sexual identity, serious disability or illness," Allen said.
But sometimes, there is no clear consensus on what constitutes hate speech because the words themselves are ambiguous, their meaning is unknown, or the context surrounding them is unclear. Under these circumstances, Facebook will decline designated publications internally, and will decide whether to delete it or not.
The giant in the field of social communication is also looking at artificial intelligence to automatically filter out hate speech, but the platform now relies on users who report such publications.
He added:
As we invest in these promising developments, we are far from being able to rely on artificial learning and artificial intelligence to deal with the complexity of assessing hate speech.
