AI Algorithms Get Tripped Up in Fighting Hate Speech

Hate speech at times slips through the finest filter, an Oxford study finds, while a more literal system can block similar terms used without ill intent.