As a community manager who has dealt with communities with incredibly sensitive topics (Mental Health communities, Fringe Adult communities, Crypto communities) I heavily support this, including a seperate log file for Admin perusal every time it is used, who used it, and to view which user.
For instance, in one community I helped manage (Which was really a large quantity of smaller, more tight knit communities), I had to create and implement a policy as we had situations where Admins were viewing messages from people in their hyperlocal community, just because. Eventually we setup a conflict of interest situation where you couldn’t deal with a case if it was someone you had dealings with, or was in your local area (large remote group of CMs)
When we enabled logging, we had to fire FOUR CMs, as it turned out that they were using their powers to effectively spy on their local community. Other CMs had admitted to doing the action also until we wrote policy against it.