The problem here is that it cannot possibly work robustly. It does what it does by trampling all over the error correction codes and using out of spec pixel shapes.
This is not the sort of problem that will go away once the technology matures. It is fragile by its very nature.
It is absolutely OK to recommend against it on that basis.
I don't think you're right about that, how could it work at all? Are you even sure how it works? I was able to scan all 21 perfectly, less than 1 second each, on iOS 18 a couple of inches from retina display. I have no idea how it works, but it seems pretty robust to me.
I know people can't do it on some Android phones, and apps. Okay. But still seems pretty good and not a fundamental flaw. Absolutely not OK to blanket recommend against trying that. Why limit anyone? Just encourage everyone to try for themselves. Get more data. Premature to make these negative recommendations on 1 test. Rather than short-circuiting further curiosity and exploration, expand it!