Rule

If you decode the QR code by hand, it reads

bo(~dsd␡`}#?j#o(oeacck█io~egnrd██nvev██~█q␓.█q␗██a␜n███'

Of course, I can't do Solomon-Reed error correction in my head so I can't recover the damaged bits. A computer won't succeed at it either - there is likely too much damage. You might be able to get a read eventually by filling the grey areas with noise and trying to scan that repeatedly (enough bits will at some point be "correct") but there is no guarantee that the resulting string will be what was intended, though if it corresponds to the above decoded bytes it probably is. It sure does not look like a URL - I am pretty sure I didn't make any mistakes as recoverable bytes all fit the 0-1----- or, more rarely, 0001---1 pattern - the latter are likely dividers of some sort. This suggests it's some kind of internal-use encoding with 6 bits per byte. Interesting, I thought it would just encode the URL and/or the bonus code (but it's not valid ASCII text so no normal reader app will read it, and you don't need 55 bytes for 6 alphanumeric characters!) Very strange given there is a URL right next to the code but the code is obviously different.

I'd really like to see an uncensored code from another such card.

This is why you always exclude vowels from randomly generated strings you hand out to users. You never know when your infinite typing monkeys will write something much worse than some Shakespeare.

Guess discount codes should get sanitized against naughty/spicy word lists. With enough logic to prevent anyone from e.g. getting the first few letters of the N word except with a 1 in place of the i.