What is nearly universally true is that people's ability to make and distinguish between phones not present in their early environment is very weak. I will probably always have a difficult time distinguishing between "cot" and "caught" because they contain allophones of the same phoneme in my dialect. Same for "merry", "mary" and "marry".

When I took a college phonetics course, I and a classful of other students more than doubled the number of distinct sounds we could distinguish and produce. So it can certainly be done. I think adults normally don't because they don't need to, since mapping to their native language's phonology is so much easier. Also, when I took a foreign language in high school, there was no IPA and none of the teachers had the linguistics training necessary to explain the cause of an accent even if they took the time to do so.

But it's certainly true that language learning is automatic for children, and not so automatic for adults.