I can confirm Edward Conway’s answer, as I was who he tested it out with.
I was surprised that moderation can not only scrutinise but edit PMs, but Nikki Primrose has just reported the same experience: I received a notification that I’d violated BNBR but the link was to a message in my inbox that I’d never replied to. What is that about? She got Benburred (h/t La Gigi) for receiving an offensive message (?!), but her correspondant got his message removed from her inbox.
My first surmise is that someone reported you, but..
… OP, this was happening while you were typing? That sounds… draconian, and unmanageable. And it certainly doesn’t sound like a reaction to a report. It does sound extraordinary that people’s PMs should be prefiltered, especially for something relatively innocuous.
I don’t know that you can get it back; but an appeal to mods is always worth trying.
Yes, Shashank, I know. 🙂