[cluedos]

Neither of those estimates seem to be correct. Per https://pubmed.ncbi.nlm.nih.gov/12476264/, the prevalence is 0.018%, or around 1 in 5,500.

That doesn't mean it's impossible to observe the sex in most of these cases though, it just takes more than a quick visual check to determine.

The really tricky cases are where the individual has reproductive organs of mixed types, particularly where it involves some sort of genetic mosaicism or chimerism. These ones are where we could reasonably say that sex is only assigned and not observed, but it's very rare. Rarest of all is where someone could be plausibly regarded as both female and male.

Generally, I think it's best to avoid the terminology of "assigned at birth", because it comes with the implication that sex can be arbitrarily reassigned. Something like "incorrectly observed" would be better, in cases where a mistake has genuinely been made.