Suddenly a lot of scammer will contact you in less than 5' with techniques that you cannot imagine are real. For example, telegram handles with the same name as the channel admin but using unicode characters to make tou think it is the same account.
The root issue is the lack of skepticism and verification. At the same time humans have limited energy and verifying everything causes significant fatigue over time, so the problem might as well be intractable.
I believe different browsers have different heuristics about when to switch to that representation - suspicious characters, mixing scripts in the same URL, and so on.
For domain names I use a "corporate" setting in Firefox, disallowing the use of DoH/DoT: to make sure that every single domain name resolution goes through my own local DNS resolver.
And my firewall inspects every packet on port 53 and rejects any packet containing "xn--" (the way they encode Unicode chars in ascii URLs).
For text: my editor is configured to display in bold, fluo, on a dark background any character that is no a visible ASCII char (except newlines and spaces) and "zero width" char are forced to have a width.
But it's a losing battle: too many people don't understand the security implication of using Unicode everywhere.
The most enraging in all this is how stupid these homoglyph/homograph attacks are to pull off: any dumbfuck can pull it off. The bar is insanely low.
Ah yes, homograph/homoglyph attacks. Bruce Schneier warned about that decades ago. He predicted all these homograph/homoglyph attacks.
Work for domain names too seen that there can be internationalized domain names.