I’ve definitely heard folks say that early output is unhelpful or can even be harmful.

I tend to think it’s somewhat beneficial. Especially shadowing (repeating or speaking along with recordings of native speakers).

At some point you’ll have to practice generating messages in order to build that muscle. But I think more important is to do lots of listening and reading, so you start to build this intuition that tells you when “that doesn’t sound right”.