Another viewpoint is that it's about privacy (e.g., unwanted tracking) and security (e.g., homograph attacks).
As LLMs are increasingly used everywhere, this provides a way to normalize text as it moves between different systems.
Are there interviews where the candidates submit large blocks of plain text? I am not aware of them. There are plenty of opportunities to cheat in coding challenges, but the unicode tricks obviously won't apply to it, the programs won't compile.
Instead, by "cheating", I mean cheating in schools/colleges, as well as passing ChatGPT output as blog post and pretending it's human-written. First one is very unfair to the other students who are not cheating, as well as to future employers who discover that their new employee can only write on the topics that ChatGPT is familiar with. The latter is simply disrespectful to readers.
And in the end, unfair to the student themselves as they won't actually know how to write for themselves - and there are occasions in life when it is critical to be able to write for yourself, by hand. Not very many, but the ones that still exist are probably not going anywhere.