I tried asking it some questions in German, and the answers read similar to auto-translated spam you often find on the internet. (If you look for stuff in German, a lot of results are often just auto-translated from English) So it seems it's already suffering from spam.

A thought just hit me: If AI researchers use text from the internet to train their auto-translators, and a lot of text on the internet is created with automatic translators, the translators will end up reinforcing their errors and develop a very distinct style, and since people read a lot of the autotranslated stuff, maybe people will also start writing in that style...