They moderate things that risk bringing bad PR, including in some cases putting overzealous rules that would cause false positives like you described.

However, it's clear they aren't acting in good faith and aren't actually interested in moderating bad content. Cybercrime-related Facebook groups which don't even attempt to conceal themselves are commonplace on Facebook: https://krebsonsecurity.com/2019/04/a-year-later-cybercrime-... despite detecting them being trivial based on a simple keyword match.