Attention to pitfalls of #Neural #MachineTranslation especially when the result sounds wonderfully fluent but it actually conveys the exact opposite meaning! Look at this example and imagine the consequences:

In a test from Greek into Spanish, the phrase "the human rights of the refugees are violated as well as the human rights of the natives" became "refugees violated their human rights as well as the human rights of the natives". #NMT #beware

Και οι πρόσφυγες, καταπατούνται τα ανθρώπινα δικαιώματά τους αλλά και τα ανθρώπινα δικαιώματα των γηγενών. >> Los refugiados también violan sus derechos humanos, así como los derechos humanos de los nativos.

The interesting thing is that the syntaxis in the source text is a bit off, but perfectly understood. The syntax of the translation is "normalized", it is absolutely correct, but the meaning is the opposite.