Memes

In many of them but not all, because it's become convention and has been enshrined in their documentation policies. cGMP just requires that your quality management system has a policy in place that specifies how to document the date, and when exceptions are allowed (for instance, data printouts where YYYY-MM-DD is often the default).

It's also the reason some labs require you to initial/date every page of printed data, and some only require you to initial/date the first and/or last page. I've seen FDA auditors be okay with both, as long as you can justify it with something like: our documentation policy defines the printout as a copy of the original data, and the original data as what's stored on machine memory with electronic signature; versus: our documentation policy defines the original signed/dated data printout as the original data. In any case, it still has to follow 21 CFR part 11 requirements for electronic records & signatures, where the only date predicate rule example they give is 58.130(e), which itself is broad and only applies to non-clinical lab studies. It's notable that the date format 21 CFR 11 itself uses is actually Month D, YYYY, with no zero padding on the day.

And if you don't have IQ/OQ/PQ documentation showing how you locked down and validated the software's ability to maintain an audit trail you can't even use electronic records (or signatures).