QR codes essentially just encode text, as long as you're using a sensible QR code reader and check any URLs before opening them there's minimal risk to scanning a QR code.
Respectfully I think this is a minimal attack vector in this case due to the limited character set of urls. But thanks for the callout, I didn't know there was a name for this sort of attack.
Modern browsers happily show you the actual characters, while sending their encoded entities to the server. So, from a user perspective there is no ASCII limitation. Case in point: söhne.at (just some random website, I have no idea what they are or if they are legitimate)
They'd still resolve via DNS to an address in ASCII though, right? Wouldn't that only be an issue if ICANN didn't have a monopoly on DNS registration? i.e what we already depend on for a semblance of convenience without totally compromising opsec
I mean, generally speaking, just don't click on random links. This is a random link. Qr codes are valuable but we're conditioning society to just be cool with clicking on random shit without putting much thought into it.
I may have in the past put lyrics from “Never Gonna You Up” or links to the music video on YouTube in QR codes I printed on blank business cards and left them in public places around town.