I’m not saying this is untrue, it may only apply to certain combinations of words or maybe even only channels over a certain size. But I’ve been chatting to a couple people over on YT about Mastodon the last few days, and nothing untoward happened to our convos.
That said, a lot of the time perfectly normal comments get flagged as spam for literally no reason at all, so it could just be general incompetence rather than malicious targeting.