MENLO PARK, Calif. (Reuters) – Facebook Inc is working on automatically flagging offensive material in live video streams, building on a growing effort to use artificial intelligence to monitor content, said Joaquin Candela, the company’s director of applied machine learning.