@Surlytom I'm unaware of any direct studies of this type, in part because I can't imagine an IRB signing off on an experimental design that guarantees exposing minors to inappropriate/harmful content.
But here's an anecdote: when we had the most restrictive filters and monitors in place, you know where our 5th graders started creating abusive chats?
Google Docs. The tool they were expected to use for classwork. And even when Google (eventually) created some measure of monitoring for that platform, the kids immediately learned how to use emoji to speak in code to avoid said filters.
They always find a way.