With all the deep learning algorithms probably running on FB's servers, how come they cannot automatically detect these crude attempts at scamming their users? It seems they all follow certain patterns - keywords ("virus", "critical system error", etc.), a general alarmist tone, and a call for action below.