I do really hope that we can prevent the knee-jerk reaction of just preventing all of UTF-8...

Having said that, what is the best way to prevent this in comments without blocking UTF-8?

"Hi. I run this place. Can you please not do the Zalgo thing all the time? It drives me crazy. Thanks!"

Or in a forum maybe even "Hi. If you want to speak Zalgo, please take it to the Zalgo subforum in the future. Thanks!"

Or look for people piling up a bunch of diacritics on a single letter and throw an error message if you'd rather have the machine do it.

¯\(°_o)/¯

I imagine cutting off "weird unicode" won't be too hard. For English, it's super-easy - just leave ASCII with letters, numbers, punctuation, etc., cut the rest. For languages with more complex characters it will require some whitelisting I guess, and smart code to disallow stacking diacritics, etc. Sounds quite possible, I wouldn't be surprised if such libraries already existed.

And then you'll not be able to quote foreign text. Ban the user if he misbehave, not the language