Both are used. What's happening here is that
data is a Latin word which has been taken into English. But while in Latin it's plural, in English it's usually treated as grammatically identical to
information, which of course is uncountable. So
data too is seen as uncountable and therefore singular:
The data on this is confusing. Some people though, who know it's a Latin plural, will use it with a plural verb :
The data on this are confusing. The plural is much rarer though - if you feed both
data is and
data are into the
Cobuild concordancer, you get just over 407 thousand occurrences of
data is and just under 199 thousand occurrences of
data are.
So whichever. And if someone tells you they "must" be plural because the original language was, ask them why they don't do the same with words referring to pasta. In Italian
spaghetti, tagliatelle etc are plural, and so you say (in Italian):
the spaghetti are ready. But in English they're treated as grammatically equivalent to pasta - uncountable and therefore taking a singular verb :
the spaghetti is ready.
As languages adopt words from other languages they frequently (almost always?) adapt them, whether in terms of meaning, in pronunciation or grammatically. You can probably think of various examples in languages you know which have adopted English words and distorted them. Or other foreign words used in English which have been distorted.